DEV Community: adalycoder

10 Machine Learning Methods that Every Data Scientist Should Know

adalycoder — Tue, 24 Mar 2020 13:18:50 +0000

Machine learning is one of AI use-cases that learn a machine to perform one or another function depending on the user request. Businesses use ML algorithms in multiple spheres: from recommendation algorithm development in video streaming to sales prediction in online retail. In a nutshell, such ML algorithms, known as models, represent the ways to apply data in the context of a business problem, and then, receive insight. They are two main strategies of teaching ML algorithm perform its functions:

Supervised machine learning, i.e., you train an algorithm using an existing dataset
Unsupervised machine learning, i.e., you need to find previously unknown patterns in data set without pre-existing labels If you consider entering as a data science field, you should be aware of the ten most popular ML methods that belong to both supervised and unsupervised techniques that data scientists currently use to manipulate data.

Naive Bayes or Classification

The classification method, based on Naïve Bayes theorem, is a probabilistic classifier that belongs to the supervised machine learning. Classification performs well to categorize the input data into predefined groups. You can apply the algorithm to assume that the presence of a particular feature in a class is unrelated to the presence of any other feature. You can use this algorithm for:

Text classification, i.e., sentiment analysis
Document categorization
Spam filtering
News classification

Regression

Regression in another ML method that belongs to supervised algorithms. The definition of regression is a data mining function to predict a number based on a set of previous data. These techniques could be used for simple linear regression to more complex regressions, such as polynomial, decision tree, and random forest regressions. Regression models have numerous use cases across the following industries:
-Financial forecasting
-Trend analysis
-Marketing
-Time-series prediction
-And even drug response modeling

Clustering

Clustering, which belongs to unsupervised machine learning methods, allows us to draw references from datasets that consist of input data without label responses. You can apply this algorithm for finding:

Meaningful structure
Generative features
Explanatory underlying processes
Or grouping inherent in a set of examples Clustering methods only apply visualization of elements to inspect the quality of the solution.

Dimensionality Reduction

Dimensionality Reduction is a method of unsupervised machine learning. In this context, ‘dimensionality’ means the number of features (i.e., input variables) in your dataset.
By using Dimensionality Reduction, you can reduce or remove the number of random variables we take into consideration by obtaining a set of principal variables. In other words, this method allows you to find a smaller set of new variables containing the same information as the input variables.

Ensemble Methods

Such methods leverage several supervised machine learning methods at once to receive a higher quality prediction compared to the forecast each model could provide on its own. Ensemble methods in machine learning are divided into two main groups:

Sequential methods that exploit the dependence between the base learners
Parallel methods that exploit independence between the base learners which reduce the number of errors by averaging The example of the ensemble method is Random Forest algorithms that apply the combination of several Decision Trees, previously trained using different samples of the data set. In this way, ensemble methods decrease variance (bagging), bias (boosting), or improve predictions (stacking).

Neural Networks and Deep Learning

Inspired by the human brain, neural networks include a set of algorithms developers that apply clustering raw input, labeling, and machine perception to recognize patterns. Such patterns could be in forms of numbers, vectors, images, sounds, and text. To achieve this, neural networks use different input, output, and even hidden layers of parameters to the model. Thanks to Neural Networks' flexible stature, you can add as many layers as necessary. Thus you can build well-known linear and logistic regression. You can use Neural Network for extracting features previously fed to other algorithms for classification and clustering. At the same time, Deep learning is a part of a neural net that includes numerous hidden layers and encapsulates a wide variety of architectures.

Transfer Learning

This machine learning method re-uses particular parts of previously trained neural net and then adapts them to different but related tasks. For example, after you trained a neural network using data for a task, you can use the same fraction of the trained layers combined with a few new layers that you can train for the new task. In this way, transfer learning significantly improve the sample efficiency of a reinforcement learning agent. You can apply transfer learning on your own predictive modeling problems using the following approaches:
Develop Model Approach
Pre-trained Model Approach

Reinforcement Learning

Reinforcement learning is a machine learning technique that enables an agent to learn from experience by trial and error, applying feedbacks from its own experience. We can compare this method with supervised machine learning since they both use the mapping between input and output. Still, unlike supervised learning, reinforcement learning uses rewards and punishments as signals of positive and negative behavior. You can apply reinforcement learning in case you have no historical data about an issue since RL does not need information in advance.

Natural Language Processing

While Natural Language Processing is not a machine learning technique, but rather a machine learning algorithm that applies artificial intelligence. This method allows understanding, analyzing, manipulating, and potentially generating human language using such texts as social media comments, online reviews, survey responses, even financial, medical, legal, and regulatory documents. In order to map the text in numerical representation, the NLP algorithm computes the frequency of each word within each text document.

Word Embeddings

Word Embedding is a way to improve the ability of networks to learn from text data. Word Embedding technique uses the setting of language modeling and feature learning techniques applied in Natural Language Processing, where words or phrases from the vocabulary are mapped to vectors of real numbers. This method is based on the idea of using a densely distributed representation for each word. Thus, word representations allow finding similarities between words by computing the cosine similarity between the vector representation of two words.

Using Social Media at Work: How Much Is Too Much?

adalycoder — Tue, 24 Mar 2020 11:11:49 +0000

According to a Digital Information World infographic, 18 percent of employees spend 15 minutes on social media during the average workday. Moreover, 19 percent of workers spend 1 hour on social media, while another 9 percent spend two hours on the social media platform of their liking.

While these statistics are not quite alarming per se, many organizations have since woken up to the fact that social media presents equally significant downsides.
This increasing concern comes even though many companies across the globe continue to praise the revolutionary impact that social media had on the workplace. After all, in the very least, social media has enabled better connectivity between employers, employees, and the rest of the world.
Challenged by the counterproductive impacts that social media has presented in the workplace, it has become necessary for companies to find the perfect balance in its use to ensure that only gain is realized. So, to help, this article explores some tell-tale signs that imply too much use of social media in the workplace. Go through them to know when it’s time to cut back on the hours you spend on your favorite platform.

When Social Media Becomes Addictive

Using social media makes people more and more absorbed as they try to keep themselves occupied with images, posts, and videos from different parts of the world. In the end, they become addicted and start performing dismally at work.
If you realize that you cannot put your phone down and take a break from social media, then there're chances that you are addicted. Your addiction is even dire if you have started becoming sluggish at work. So, to get back on the right track, it helps to get some help in the form of counseling and so forth.

When Social Media Can Harm Your Privacy

If you have reached the point of oversharing your personal information such that the details you've given out there can be used to harm you, then you've probably had too much social media. When in this situation, it helps to step back from an app like TikTok, or any other, where your information is present. This way, you can reclaim your life slowly and win back your privacy.

When You Crave Attention

Tracking your posts and photos often is a sure sign of too much social media. This habit is of a person who has become an attention-seeker and social media-dependent to the point where they live for the attention that they gain from the various platforms they are on.

When You Start Having Psychological Issues

If you realize that you get an immense psychological boost when your posts go viral and feel miserable when that does not happen, you've had too much social media. It is a fact that psychological problems are common for people who tend to peg their self-worth and importance to others, in this case, when others like your posts.

When It’s Affecting Your Productivity

Using social media at work from time to time is not bad. If brief social media breaks help you to reduce stress, increase morale, and refresh your mind, it might be good for productivity as well. However, if you spend most of your working hours on social media such that it affects productivity, remember that it's your reputation you are putting on the line. If that is the case, then you are most likely overdoing social media.

When You Are Using Social Media to Vent About Work

We are often told that it’s not professional to complain about your boss or workplace to others. That includes social media too. However, today the majority of people use social media platforms to share their happy moments as well as vent about their frustrations. If you are using social media to complain about your overbearing boss, post drama going on in your office, you probably need to take a step back and evaluate the consequences of your actions. It’s not just your current employer who might be unimpressed about such actions but also future employers.

When It’s Hurting Your Relationship with Other Employees

Jealousy among peers doesn't happen with teens alone. Working adults also struggle with it. Employees who become envious of workmates who have more success can create discord in the workplace. If social media is straining relationships between colleagues, creating bad blood withing teams, promoting bullying and other types of harassment, it's most likely that it is being used too much. If the situation is not regulated, it can lead to lower overall productivity in the company.

How to Avoid Getting Consumed with Social Media at Work

Overdoing the use of Facebook, Instagram, Twitter, etc. at the workplace can have dire consequences on relationships, productivity, and happiness in the workplace. If you feel like you are using too much social media, especially during your working hours, here are some tips you should try:

Turn off social media notifications: if your social media notifications are on, you will most likely be distracted by new messages every few minutes. Turning them off will, therefore, help you stay away from your phone for longer.
Be discreet: be careful about what you post on your social media accounts. Avoid sharing ill-tempered posts about your workplace, colleagues or boss, or other sensitive workplace matters on social networks.
Set a Limit: decide how much time you want to be spending on an app and stick to that limit. To make that possible, however, you should download a social media tracker or app to help you follow the plan.
Keep your smartphone away: try keeping your phone away from your desk. That way, you can concentrate on your work better without constantly looking at your phone screen.
Delete the addictive app: if everything else fails at deterring you from overusing social media, consider deleting the apps for a week or more to overcome the addiction. ##Bottomline Social Media is more like a down-edged sword. If used responsibly, it can have positive impacts on employees’ emotional and psychological wellbeing. However, if it is used too much, it may lead to undesirable consequences as well. So, assess yourself to know which category you fall in and take the necessary remedial measures.

Tech Stack for Taxi App Development: A Useful Guide for Developers

adalycoder — Mon, 02 Mar 2020 16:24:34 +0000

A tech stack is a combination of programming languages, tools, and frameworks used to build websites and applications. It occurs when a layer of an app (front end or back end) is made atop the another using codes and hardware modules. When developing a taxi application, you should choose the best taxi app technology stack. Also, you should stay flexible and be ready to change your tech stack over time. For example, Uber, a successful on-demand taxi app startup, has changed its tech stack since it’s inception. That said, here are the technical aspects for building a taxi application that is sustainable, scalable, and secure.

Geolocation Technology

Location services are an integral part of any taxi application. They enable customers to locate taxis and drivers to find their passengers. To develop a device’s location feature for your app, you will need Core Location Framework for iOS and Google’s Location APIs for Android. As far as navigation is concerned, there is MapKit for iOS and Android Google Maps Android API for Android.

Notifications: Email, Push, and SMS

Frequent alerts for both drivers and passengers are essential for any taxi application. Today, these alerts can either be push notifications, SMS, or email. If you are developing for Android, you will need Google Cloud Messaging to enable push notifications. And, for iOS, the tool to use is Apple Push Notification Service.

Inter Application Communication

Taxi applications should be connected to allow communication between drivers’ and customers’ devices. Two apps can communicate using Wi-Fi or mobile data, and make it possible to perform such operations as booking a taxi, determining the location, route monitoring, pricing, chat messaging, etc.

Cashless Payment System

Payment is perhaps the most crucial aspect of an Uber tech stack. The most popular card payment gateways for cab apps are Stripe (used by Lyft) and Braintree (used by Uber). These payment services are convenient for taxi app startups because they support multiple currencies, cryptocurrency (Bitcoin), split fare, and a variety of payment options (PayPal, American Express, Apple Pay, Android Pay, etc.).

Storage & App Infrastructure

For the front-end and back-end services to interact and coordinate efficiently, taxi app developers must put together a fully-functioning ecosystem. Database systems like SQL Server, MySQL, PostgreSQL, etc. can be used to store app data. Tools for other development processes include:

• Inventory management: Clusto
• Searching & visualizations: Logstash
• Distributed storage: Hadoop Warehouse
• Caching & Queuing: Twenproxy
• Long-term data storage: Riak, Schemaless

Website & Mobile

You will need to build on-demand web and mobile taxi apps to meet the different needs of internet users. Each web app will require a database, client-side, and server-side. The best languages to select here are Java, CSS/HTML, Python, or Node.js. The programs and tools you will need for each platform will differ as below:

For iOS:
• Backend: MySQL & Schemaless
• Storage: LevelDB
• Repository: Buck
• Crash detection: KSCrash
• Testing: OCMock

For Android:
• Architecture: RxJava
• Automation coding: Expresso
• Logging: Timber
• Image loading: Picasso
• Networking: Gson, OkHttp

Final Words

When you implement all the mentioned components and use the right programs, languages, and tools, you will get a great site or app for on-demand taxi services. Clarify the design and requirement for the app you are about to create, choose a convenient coding language, and use this stack as a checklist.

MVP for a Dating App: What to Include

adalycoder — Thu, 16 Jan 2020 11:55:59 +0000

Dating app development became a trend today. More and more people are using dating apps to meet someone special. Online communication is a very popular way to start romantic relationships, and by creating a new unique dating app, you will help thousands of people find their soulmates, as well as make it a profitable business for you.

Once you have explored your potential competitors, defined your target audience, and learned about the potential users’ needs, it’s time to create an MVP with the main features that will define your application. In this article, you can find helpful tips on what functionalities to include in your dating app to come up with an effective service that will gain popularity.

Social Login

It is the first thing to add to your dating app. We recommend using social login to avoid long registrations that often push people away. Make a quick sign-in through Facebook, Instagram, or Google account.

User’s Profile

A user’s profile is a core place where a person can upload their photos, indicate hobbies, tastes, interests, and tell more about themselves. An option to extract all the info from their social profiles will save users’ time.

Geolocation

By allowing to indicate one’s location, you will help soulmates to agree on where to meet up easily. Integrate a tracking system to let people nearby meet each other.

Matching Algorithms

Smart AI algorithms can match users’ profiles according to their common interests, hobbies, and tastes. Make the searching process fast and personalized.

Chat

Create a chat to allow people to communicate easily, exchange pictures, videos, various stickers, or catchy GIFs.

Notifications

Let your users see important notifications from the app. For example, if they received a message or have matched with a person they liked. People will visit your app more often, and the chances to meet a soulmate will be much higher for them.

Swipe Feature

It is the simplest and most useful feature you can add to your app so that users could show their preferences with a single swipe. For instance, you can allow them to swipe left if they don’t like a person or right to see if the interest is mutual.

Settings

Allow users to customize their account settings according to their needs. Let them change the font size, background color, or sound.

Admin Panel

This core feature should be available for PC and hidden from all users. An admin panel should be designed for your team to resolve any conflict situations that occur, block people, and be in control of everything.

These are the essential features that you should include in your dating app. Though, you can experiment and add some new and unique options that will set your product apart from the competitors. Also, make sure to come up with a catchy name and create a great UX/UI design.

Final Words

Dating apps become more and more popular nowadays. So, if you have decided to launch another fantastic functional app, including these basics is integral for your success. Don’t be afraid of experimenting and adding something new to make your app stand out. However, remember to get the essentials that will make your app work smoothly.

What Is a Developer Evangelist?

adalycoder — Wed, 11 Dec 2019 08:57:40 +0000

Evangelism is simply spreading a particular message to a target audience. For example, in marketing, evangelism is a form of word-of-mouth marketing whereby companies gain customers who believe strongly in the brand. Therefore, they also persuade other people to purchase and use the brand’s products or services. From this definition, it is fairly easy to understand who a developer evangelist is.
According to the Developer Evangelism Handbook, a developer evangelist advocates, translates, or mediates between a company, its technical team, and external developers. Developer evangelists/advocates are usually people with technical experience. That is, they must be developers themselves. They should also have strong networking and communication skills, as their job is to create awareness about particular technologies to different audiences and get their support.

Why Is a Developer Evangelist Important to a Software Company?

Developers are usually skeptical about things that don’t relate to tech. That is why non-tech people are not the most suited candidates to sell software products to programmers. A developer evangelist fills in the gaps that a PR or marketing professional cannot. That is, this person has to convince not just customers, but also developers to try out the software, and, as a result – spread the word. So, a developer evangelist must also be the one who can package and deliver the message in the most effective way.

What Is the Job Description of a Developer Evangelist?

So, what does a developer evangelist do? If you want to get back the cost to develop an app and make a neat income from it, you will have to address a developer advocate/evangelist. The main job of these specialists is communication. They build a critical mass for a digital product and establish it as a technical standard.

While professionals in this role spend some time in the actual development process, the developer evangelist role is mainly to assist different departments within the companies to communicate efficiently as well as create awareness about the software on the market. Their responsibilities involve a combination of programming, customer interactions, and content production. They often work together with technical content creators, marketing teams, client support, and community managers.

If you’re in this position, be prepared to work outside normal working hours, adjust your schedule to include traveling commitments, work on the go, and start conversations with all types of people. Among the main tasks you will perform are:

• Handling the company’s social media interactions
• Take part in public discussions
• Make training content, best practices, and video tutorials
• Bring in and foster relationships with influencers
• Participate in developing software products
• Follow up on market trends and the competition
• Create sample codes
• Share open source codes on developer channels
• Get involved in other developers’ open source software
• Organize and take part in workshops and other relevant events
• Use customer feedback and data to foster growth

How Much Does a Developer Evangelist Earn?

Jobs in technology are among the highest-paid today. So, it’s no surprise that a developer evangelist salary is high. According to Glassdoor, the average base salary for this position in the US is $122,550 a year. Average additional compensation is about $10,926. Going by another source, ZipRecruiter, the average pay is $114,159 annually.

How Can You Become a Developer Evangelist?

There is no specific university program that will train you to become a developer evangelist. The requirements for a developer evangelist job is different depending on the company. However, if you are already working in software development, are passionate about making people understand the products of your company better, and are a good communicator, you can try your hand in this path.
That said, remember that this position is not entry-level. The majority of developer evangelists you see today started out as programmers before moving on to the position they are now. If you are already a developer, you can start shadowing people in this role or start working part-time to prepare yourself to ease up to the job later.

Final Thoughts

A developer evangelist is both an ambassador for the company and a customer support representative. It means these specialists advocate for and encourage people to engage with or use particular products. They also offer valuable information and answer customer questions. Their experiences and expertise benefit almost every department in a company, from software production to marketing.
As with any other job out there, this role also has its challenges. For example, you will need to be approachable and enthusiastic at all times. You may also need to travel a lot to conferences, meetings, and events related to the brand. However, if you are passionate about software development, like to talk and inspire people – this job is very rewarding.

Basic Features Every Social Media App Should Have

adalycoder — Mon, 18 Nov 2019 09:39:53 +0000

In this age, social media is an essential part of the internet. Some people even view it as something that depicts the present state of the internet in general. Do you remember the last time you used your mobile phone or PC? There are high chances that you have visited one or several popular social media applications, for example, WhatsApp, Facebook, Twitter, Snapchat, YouTube, or Instagram.
When it comes to developing social media applications, the first thing you have to do is to understand the fundamental elements that make them up. What functionalities do users want? What aspects do social media giants share? While the answers to these questions are not that straightforward, the bottom line is that leading social media applications have some shared characteristics that each aspiring developer needs to adopt in his app.

Typical Features of Social Media Applications

Below are some of the features that are shared across various social media platforms. Most successful social media apps have a number of characteristics in common.

Authorization

Making the authorization of users straightforward to ensure their smooth onboarding experience is one of the essential features that social media applications share. All told, the vital thing is to ensure that there are as minimal steps to authorization as possible.
Creation and Editing of Profiles
Another feature that is typical for various social media apps is the possibility of not only creation but editing of user profiles. Such applications allow users to create and edit their profiles on the specific app so that others could find them there.

In-App Messages

Social media applications tend to have in-app messaging. This functionality is common practice for anyone who knows how to create a social media app and a vital step that developers must put into account as a way to satisfy the users’ requirements.

Photos and Videos Uploading

A feature that is highly on-demand in social media apps is image and video uploading. Regardless of the type of social media network you are looking at, there’s always the option for uploading a photo to one’s profile and, in some cases, editing as well.

Connecting with Other Social Media Networks

Users want to share their images, content, and videos across various platforms. As such, many social media applications offer users a way to link and share their content on multiple social media platforms.

Search Bar

Including a search button in the app helps users to look up for people, communities, pages, among other things. This functionality has become quite common for social media applications.

Push Alerts

Push notifications are not new to social media platforms. In fact, they are commonplace in the mobile app development world, no matter of the genre of the application.

The features mentioned above are the most common across different social media apps. However, as someone who’s looking to develop your own social media platform, you will need to think beyond these and, at the same time, about how your app will make an impact in the social media world. This foresight will help you to create an app that has all these features plus plenty of other functionalities.

COST SAVINGS THROUGH UNIT TESTING (Article for Beginners)

adalycoder — Fri, 20 Sep 2019 14:58:55 +0000

Recently we worked with a client that was interested in creating an instant messaging app. They wanted this application to be usable across the major mobile platforms (Android, Apple, and Microsoft). Using unit testing we created this application quickly and cost effectively. Implementing unit testing on projects is an excellent way to help a project budget.

What is unit testing?

Unit testing is the programming process to validate the correctness of the individual modules of the program source code. The idea is to write tests for each non-trivial function or method. This allows you to quickly check to see whether the resulting change in the next code to the regression – that is to errors in the field – is of tested programs. This facilitates the detection and elimination of such errors.

In practical experience, when applied to a variety of projects, it is safe to say that for every 1 hour spent creating a unit test, 100 hours of manual testing time are saved. This process helps to keep software working during the agile development process and for every milestone thereafter. There is no new technology being built; it is simply optimizing the methods that already exist.

Many developers do not like to use unit tests because they feel that they have coded perfectly – however the fact is, humans do make errors. By using unit testing, errors are easily caught. Human interaction is still necessary to make sure things are running smoothly and for a second look – in other words, for oversight. But unit testing saves a very large chunk of time.

What happens when there is a bug in a released application?

Every kind of methodology will produce some kind of mistake in the software, but when a unit test has been created we can detect this mistake by simply clicking the unit. In five minutes the results of the testing will be available. The developer almost immediately knows what function doesn’t work, how to fix it, and how to save time in an update. This is very important when working with mobile applications. The longer it takes to fix and respond to a bug in an application, the fewer people will be purchasing and using the application.

Unit testing can take more time on the front end – however the time and cost savings over time make this methodology very desirable.

5 Tips to Achieve Minimalism in Your Web Design

adalycoder — Wed, 21 Aug 2019 08:35:06 +0000

The present online trend has a simple design that makes a website look attractive and appealing. This type of designs is known as a minimalist one that can be informative for every user who visits the website. The following are the five tips that can be used to create a minimalist website design.

Tip 1. Minimal Clutter and Maximum Focus

Many new websites have their focus on a mix of many things and they even cover up the unimportant elements and information. This kind of problem mostly happens with the first time designers who would like to show their skills by displaying many widgets and other things on the web page. Identify the things that every visitor would search for in a website and work on it to create a minimalist website.

Tip 2. Stay on the Upper Fold

Many people have the habit of not scrolling down the page to read the whole content. The visitors would always like to read the content on the first page of the website and above the fold. The web designers can improve the overall effectiveness of a website by keeping all the main elements the most needed call-to-action above the fold on a web page. Reduce the height of the header by inserting the sign-up forms and the buttons above the first fold in your web page.

Tip 3. Bring down the Page Count

If you want your website to be an informative one, then you must bring down the page count so that the visitor would not confuse with the number of pages in your website. Too many pages in your website can mislead the visitors. While writing down the content for a website, always write from your visitors’ point of view, so that you can stick to the point and write the main content short. Also a paragraph of content should be in five or six lines, so that the visitors can read them quickly.

Tip 4. Minimal colors for the best quality

The most common mistake that many new designers make while creating a website is the colors. The main key to create a quality design is to limit the colors in your website. If you are about to create a texture, then make use of the same color in different shades. Making use of too many colors on a website can hurt a human eye and the visitors can easily get distracted. Always choose two main colors that can be added everywhere in your website. If you are about to bring in a third main color to your website, then you are about to distract the visitors’ eyes from reading the content on your website. Follow different shades of the two main colors to create a minimalist website design.

Tip 5. The 80/20 rule

The 80/20 rule can help you to simplify the overall website design and pushes the focus on the main content of your website. Identify the content on your website that has the focus on the topic so that the visitors can easily read the content without getting distracted.

Creating a great workplace for developers

adalycoder — Fri, 16 Aug 2019 09:28:29 +0000

Yesterday, I came across a post about workplace experiments, which struck a chord with some of the things we do at our company to make it a great place to work for our team.

We don’t call what we do “workplace experiments”, as it is just standard practice for us. I thought it would be good to list out all the unique things that we do to enhance everyone’s jobs.

4-day week

One of our founding principles was the 4-day week. With the exception of client demands (as we still do some consulting), we regularly work 4-day weeks. No one is required to come into work on a Friday if they do not want to. The problem we are facing is keeping our team away on a Friday! The reality is, if you create the right environment and give people a goal to work towards to, you will have a tough time keeping them away from work.

Expenses

Although we do not offer everyone a company card, we never say no to anything that someone in the team wants (that is reasonable). Books, software and conferences are paid for by the company as long as we can afford them. This ensures our team gets what they want without spending the time getting them; allowing them to focus on getting on with their jobs.

Working from home & Flexible hours

fgeOur working from home policy is completely flexible and anyone from our team can work from home at management’s discretion. So whether you are having that brand new laptop delivered home or you need to see a doctor miles away, you can rest assured that it will be stress free.

Similarly, our flexible hours are actually flexible. When most outsourcing companies use the word “flexible hours” they mean that you can come in an hour later but have to stay back an hour later too! When we say “flexible hours”, we are saying “we appreciate there will be times when you cannot get into work on time or want to leave early and we are fine with that!” There is no need to stay later or come in earlier to “make up the hours”. I have yet to see this being abused at the lab!

Profit sharing

Unfortunately we don’t have a bonus scheme as we feel that it depends too much on the whims of the person deciding the bonus and offers no real method of measuring contribution.

We have a flat profit sharing scheme. It works on the basis of setting aside a share of our profits every year and dividing this equally between all team members. No matter what your role, junior or senior, director or developer, you get an equal share of the profit. Of course, there might not be a profit to share but that means no one gets an incentive. We feel this is a great way to reward overall performance of the team and company while ensuring no one feels singled out.

Overtime

Although we frown at overtime for our internal projects, work for clients sometimes dictates that we are forced to ask some of our team to work overtime. We accept this as a by-product of service work but instead of fighting against it, we embrace it by innovating on how people are paid for overtime. Instead of paying overtime rates of 2x or 3x times someone’s hourly rate, we pay them the whole amount that the company makes for that work from the client. You heard that right!

These are just some of things we try and do to make life better for everyone in the team. I would be interested in finding out what works for other people and welcome any debate on our approach.

Approach to software development, iterative/incremental or waterfall?

adalycoder — Mon, 05 Aug 2019 10:18:29 +0000

In their pure forms, iterative/incremental and waterfall models are on the opposite ends of software development. In the first approach, it is in the short, repeated cycles (iterations) which involve needs analysis, programming, unit testing and deployment of components that the final product is achieved. It is a process broken down into a number of small portions (increments) before it concludes in ultimate functionality. The second approach is not cyclical, but linear in nature. Every step of software development life cycle, from requirements gathering to production, is carried out only once and leads sequentially to following steps.

In reality, few companies adhere to either of these models in their extreme versions, going more often than not for hybrid or modified approaches. It should not be surprising considering that both iteration and waterfall have their limitations.

Iterative software development is often criticized for allowing developers to fast-track their early-stage concerns, at the level of requirements or design, and proceed without having a clear vision of where they want to arrive. This means less discipline and more chaos, which constitutes a greater burden on key resources, such as time and manpower, as issues that were under-invested at the beginning hit back with a multiplied force. Importantly, when left unresolved early on, they tend to complicate themselves over time. Iterative methods also tend to be less principled when it comes to documentation, which means that plenty of knowledge might be lost in the process.

On the other hand, critics see waterfall software development as too rigid to effectively adjust to demands of many projects, especially if it is hard to predict their full scope/functionality or if they are not very stable by nature. As software products become more and more complex, it is simply hard to predict all problem area at the start of the project. The other thing is that waterfall does not allow teams to learn much along the way – it makes it a hassle to gather practical feedback about particular components of the final product and enhance it by implementing regular corrections. Finally, in waterfall models commitment as to the scope of the tool (made possible by small-size iterations and plenty of feedback) is delayed, giving decision-makers time to choose the most adequate, most market-oriented solution.

Clearly, there is a lot of potential to integrate both approaches into something of a middle road that takes the best practice from waterfall and iteration. The mix could include discipline and documentation integrity that characterizes classic waterfall software development, as well as concentration on conserving resources by doing things right early on. The iterative approach could pitch in by bringing its massive learning value (from processing feedback) and increasing responsiveness to changing business needs when development is still in progress.
You might also read about dedicated teams in software development.

What future for Web APIs

adalycoder — Thu, 01 Aug 2019 14:20:14 +0000

Charles Darwin, the author of theory of evolution, would find ample scientific material by researching the development of Web APIs. In response to the demands of the web application market, they evolve at an incredible speed and only the fittest can adapt and be successful with developers. As competition among service providers, from cloud computing to social networks to other trending solutions, intensifies, the quality of APIs they will open up for application developers and their adaptive instincts in this respect are likely to determine interest from outside developers. Failure to attract them can quickly translate in a failure of a business to take off.

Mashery, an API management company, has conducted a survey of IT professionals asking what makes an attractive application programming interface. It is a handy how-to guide for API designers and a foretaste of which direction developer interfaces can float in the future, because there is no doubt that they are here to stay for longer.

Interestingly, the first position on the list went to good documentation, easy to follow and learn from. Respondents gave examples of such APIs as those published by Amazon, Twitter or Google. In a way, this contradicts opinions of those pros who would love to see less documentation and more code samples with their deployment in context that can be explored for guidance on further implementation. In fact, it is not hard to imagine a compromise of sorts between these two approaches.

Stability was the second primary concern, as developers expect their applications that are built around external APIs not to experience major breakdown as things change on the provider’s side. Static APIs make it possible to plan long-term engagement and build business models that reflect the stability of the system.

The survey identified mobile web application development as the area of greatest activity, with outsourcing developers seeing the easiest pathway to making a profit from their products there. Hand-held devices like tablets are a hot area too, with most companies wanting to be present in this market as soon as possible.

Mashery survey and ongoing developments in the business of web applications show that APIs have allowed developers to fit better between customers and businesses with their services. It is a three-way relationship in which everyone seems to benefit: customers by getting new tools and solutions, developers by benefiting from their apps and API providers by extending the reach of their business as their assets are put to work in new, creative ways practically without their intervention. Well, the only intervention they need to secure is building a functionality that outsiders want to use and sharing a well-designed API with outside developers.

Top 3 Natural Language Processing Libraries

adalycoder — Mon, 08 Jul 2019 12:37:33 +0000

Natural Language Processing (NLP) is the technology that powers chatbots, voice assistants, predictive text, among other text/speech applications that we interact with daily.

What Is an NLP Library?

An NLP library is a toolkit that processes unstructured data from different places allowing us to understand it and gain valuable insight. In this article, we review the top 3 NLP libraries to ensure that you settle on the right one:

Natural Language Toolkit (NLTK)

Natural Language Toolkit (NLTK) is arguably one of the most comprehensive NLP libraries out there. It implements virtually any component of natural language processing you want, for example, stemming, tokenization, classification, parsing, semantic reasoning, and tagging. Moreover, often, there is more than one implementation for each component, thus enabling you to pick the exact methodology or algorithm you want to use. Apart from that, it supports several languages.

Nonetheless, given that NLTK represents data as strings, which is excellent for simple constructs, it makes it challenging to implement some advanced functionalities. Apart from that, documentation on the library is quite dense - but you can still use the NLTK book to find your way around. Another notable downside is that, when compared to other tools, NLTK is a little bit slower.
With all that said, NLTK remains an excellent library for exploration, experimentation, and applications that require a specific combination of algorithms.

TextBlob

An extension of NLTK, you can use TextBlob to access several of NLTK’s functions in a simple way. Furthermore, TextBlob comes with functionality from the Pattern library. If you are a beginner, this is an excellent tool to use and can help in the production of apps that do not require to be overly functional.
That considered, TextBlob is used widely and is right for small projects.

SpaCy

NLTK may become slow and cumbersome when handling more complicated business applications. SpaCy is a step-up from NLTK as it gives users a much smoother, quicker, and more efficient experience. This open source natural language processing library is purposely developed for business activities such as comparing product profiles, client profiles, and text files.

SpaCy is an excellent choice when conducting market research and collecting insights because of its aspect-based sentiment analysis, named-entity recognition, and optimization of conversational user interface abilities. This library also works well with word vectors such as doc2vec and word2vec, which is not possible in CoreNLP or OpenNLP.

Overall, SpaCy stands out from other NLP libraries like Stanford, Apache OpenNLP, and CoreNLP in that all of the functions are combined to create ready building blocks. Therefore, you save time by not needing to pick individual modules by yourself. SpaCy is best used in data extraction and analysis, text summarization, and sentiment analysis.

Final Word

NLP libraries are used in text data analysis to deriving useful business insights. However, there are many options out there. Making the right decision about which NLP library is best for your projects and business is about knowing which options are available and how they compare to each other. More NLP frameworks you may find in "NLP Tools review". Choose from these top-of-the-class libraries for more successful projects.