DEV Community: Prajeen

Top 10 Reasons Why 87% of Machine Learning Projects Fail?

Prajeen — Wed, 30 Sep 2020 17:36:49 +0000

We see news about Machine learning everywhere. Indeed, there is a lot of potential in Machine learning. According to Gartner’s predictions, “Through 2020, 80% of AI projects will remain alchemy, run by wizards whose talents will not scale in the organization” and Transform 2019 of VentureBeat predicted that 87% of AI projects will never make it into production. Why is …

Top 10 Reasons Why 87% of Machine Learning Projects Fail? Read More »

The post Top 10 Reasons Why 87% of Machine Learning Projects Fail? appeared first on Break Down Data.

Top 10 Reasons Why 87% of the Machine Learning Projects Fail?

Prajeen — Fri, 18 Sep 2020 00:01:01 +0000

Machine Learning, Opinion

A lesson on the shocking reasons for your AI adoption disaster.

Photo by Jonathan Borba on Unsplash

We see news about Machine learning everywhere. Indeed, there is a lot of potential in Machine learning. According to Gartner’s predictions, “Through 2020, 80% of AI projects will remain alchemy, run by wizards whose talents will not scale in the organization” and Transform 2019 of VentureBeat predicted that 87% of AI projects will never make it into production.

Why is it like that? Why do so many projects fail?

Not Enough Expertise

One of the reasons is that the technology is still new to a large audience. In addition, most of the organizations are still unfamiliar with the software tools and the required hardware.

It seems that today, anyone who has worked in data analytics or software development who has done some sample data science projects are labeling themselves as data scientists after taking a short course online.

The fact is that experienced data scientists are needed to handle most of the machine learning and AI projects, especially when it comes to defining the success criteria, final deployment, and continuous monitoring of the model.

A disconnect between Data Science and traditional Software development

A disconnect between Data Science and traditional Software development is another major factor. Traditional software development tends to be more predictable and measurable.

However, Data science is still part-research and part-engineering.

Data science research moves ahead with multiple iterations and experimentation. Sometimes, the whole project will have to loop back from the deployment phase to the planning phase since the metric that was picked is not driving user behavior.

Traditional Agile based project deliveries may not be expected from a Data science project. This will cause large scale confusion for the leader who has been working with clear deliveries at the end of each task cycles for normal software development projects.

Volume and Quality of Data

Everyone knows that the larger the dataset, the better the prediction from the AI system. Apart from the direct implications of the higher volumes, as the size of the data increases, a lot of new challenges arise.

In many such cases, you will have to merge data from multiple sources. Once you start doing it, you will realize that they are not in sync many times. This will result in a lot of confusion. Sometimes you will end up merging data that were not supposed to merge, which will result in having data points with the same name but different meanings.

Bad data at best will produce results that aren’t actionable or insightful. Bad data can also lead to misleading results.

Source

Labeling of data

The unavailability of labeled data is another challenge that stalls many of the machine learning projects. According tothe MIT Sloan Management Review,

76% of the people combat this challenge by attempting to label and annotate training data on their own and 63% go so far as to try to build their own labeling and annotation automation technology.

This means that a huge percentage of expertise of those data scientists are lost for the labeling process. This is a major challenge for the effective execution of an AI project.

This is the reason many of the companies are outsourcing the labeling task to other companies. However, it is a challenge to outsource the labeling task if it requires enough domain knowledge. Companies will have to invest in formal and standardized training of annotators if they need to maintain quality and consistency across datasets.

Another option is to develop their own data labeling tool if the data to be labeled complex. However, this often requires more engineering overhead than the Machine learning task itself.

Organizations are Siloed

Data is the most important entity of a machine learning project. In most organizations, these data would reside in different places with different security constraints and in different formats — structured, unstructured, video files, audio files, text, and images.

Having these data in different places in the different format itself is a challenge to handle. However, the challenge doubles when the organization is siloed, and responsible individuals are not collaborating with each other.

Photo by Dmitry Demidov on Pexels

Lack of collaboration

Lack of collaboration between different teams such as Data Scientists, Data engineers, data stewards, BI specialists, DevOps, and engineering, is another major challenge. This is especially important for the teams in the engineering scheme of things to Data science since there are a lot many differences in the way they work and the technology they use to fulfill the project.

It is the engineering team who is going to implement the machine learning model and take it to the production. So, there needs to be a proper understanding and strong collaboration between them.

Technically Infeasible Projects

Since the cost of Machine learning projects tends to be extremely expensive, most of the enterprises tend to target a hyper-ambitious “moon-shot” project that will completely transform the company or the product and give oversized return or investment.

Such projects will take forever to complete and will push the data science team to their limits.

Ultimately, the business leaders will lose confidence in the project and stop the investment.

It is always best to focus on a single, achievable project with the proper scope and target a discrete business challenge.

Alignment problem between Technical and Business teams

Many times, ML projects are started without a clear alignment on expectations, goals, and success criteria of the project between the business and data science teams.

These kinds of projects will forever stay in the research stage itself because they never know if they are making any progress since it was never clear what the objective was.

Here, the data science team will be focused mainly on accuracy, whereas the business team will be more interested in metrics such as financial benefits or business insights. In the end, the business team ends up not accepting the outcome from the Data Science team.

Source

Lack of Data Strategy

According to the MIT Sloan Management Review, only 50% of large enterprises with more than 100,000 employees are most likely to have a Data strategy. Developing a solid data strategy before you start the Machine learning project is critical.

You need to have a clear understanding of the following as part of Data strategy,

The total data you have in the company
How much of that data is really required for the projects?
How will the required individuals have access to these data, and how easily those individuals can access them?
Specific strategy on how to bring all these data from different sources together
How to clean up and transform these data.

Most of the companies start without a plan or don’t start thinking that they don’t have the data.

Lack of Leadership support

It is easy to think that “you just need to throw some money and technology at the problem and the result would come automatically”

We do not see the right support from the leadership to make sure of the needed conditions for success. Sometimes business leaders do not have confidence in the models developed by the data scientists.

This could be because of the combination of a lack of understanding of AI of the business leader and the inability of the data scientist to communicate the business benefits of the model to the leadership.

Ultimately, leaders need to understand how Machine learning works and what AI really means for the organization.

For more such articles,
Why is Tuple faster than List and when to use List

Why is Tuple faster than List and when to use List

Prajeen — Sat, 25 Jul 2020 17:59:55 +0000

Let us make some inroads into uses, advantage of lists and tuples

Photo by Julia Joppien on Unsplash

What is a List

List in python is simply a collection which is ordered and changeable. Lists can contain multiple datatypes . It can be created by putting the elements between the square brackets.

myList = ['mango', 'apple', 'orange']

What is a tuple

Much like list, tuple is also a collection of ordered elements that can contain elements of multiple datatypes. However, tuple is a immutable. This effectively means that you cannot edit or delete an element of a tuple. Tuple can be created by putting elements between round brackets.

myTuple = ('Red', 'Black', 'White')

There is a common perception that tuples are lists that are immutable. However it is not completely true.

Why should lists be used for homogeneous data and tuple for heterogeneous data?

Is this really true? is this a guideline? no.. no.. in reality both of them are heterogeneous collections. You are free to use tuples for homogeneous data and lists for heterogeneous data. It is more of a culture than a guideline.

Main reason why list is preferred for homogeneous data is because it is mutable
If you have list of several things of same kind, it make sense to add another one to the list or take one from it. You are still left with a list of same things

But why is tuple different?

If you have a set of heterogeneous elements, most probably the collection has a fixed structure or ‘schema’.

#schema of tuple => (person name, age, weight)
StudentDetails = ('Job John', 35, 72)

In other words, tuples can be used to store records — related information that belong together
In such cases, tuple lets us “chunk” together related information and use it as a single entity.
Since tuple is immutable, it can be used as key for dictionary

Why is tuple faster than list?

Program execution is faster when manipulating a tuple than for a list of same size. However, it is not noticeable for collections of smaller size.

Whats the problem with list, anyway?

List has a method called append() to add single items to the existing list. So, for most of the append to be fast, python actually create a larger array in memory every time you create a list — in case you append.

This way when append is called, it doesn’t have to recreate the list.

Thus, making a list of five elements will cost more than five elements worth of memory.

What does tuple do?

In contrary, since tuple is immutable, it asks for an immutable structure. This way tuples are more explicit with memory.

Thus, making a tuple of five elements will cost only five elements worth of memory.

Finally, this overhead with memory for list costs its speed.

If you want to define a constant set of values and the only thing you want to do with it is to iterate through them, then use tuple than a list.

For more such articles,
Top 10 Reasons Why 87% of the Machine Learning Projects Fail?