DEV Community: Kevork Keheian

How to deploy streamlit app to Heroku

Kevork Keheian — Mon, 20 Sep 2021 09:00:44 +0000

Hello everyone,
This is a step by step tutorial about how to deploy your Streamlit app to Heroku.

What is streamlit?

Streamlit is an open-source python framework for Machine Learning and Data Science teams and individuals.
See it yourself: https://streamlit.io
You can deploy your app using streamlit sharing, but the problem is that you only have 3 apps with the free version.

So what you will see now is how you can deploy your app to Heroku

Requirements:

python installled
streamlit installed (You can install streamlit using pip install streamlit)
GitHub account

Once the above are installed

You can check their documentation about creating you app.

And now it's time to deploy 🎇

You need to add 2 additional files to your project.
1- You need to add a setup.sh file which is used to add shell commands inside the file.

mkdir -p ~/.streamlit/
echo "\
[server]\n\
headless = true\n\
port = $PORT\n\
enableCORS = false\n\
\n\
" > ~/.streamlit/config.toml

Then add a new file called Procfile and inside it paste the following code:

web: sh setup.sh && streamlit run app.py

And finally a requirements.txt file

streamlit
matplotlib
numpy
pandas

and the other packages that you are using in your app.

Now head over to GitHub and create a new repository and publish your code to github. You can check this article to see how you can do that.

Once the code is on GitHub head over to Heroku and create a new app:

Choose a name to your app and click "Create app"

You can leave the pipeline empty for the time being and choose "GitHub" as your deployment method.
And in the Connect to GitHub section choose your username/org, type the name of the repository and click search. After you click search you can see the repo, click on the Connect button next to the repo.

Once you click connect you can see at the end of the page The Manual deploy section. Just click Deploy Branch and watch the magic happen.

Once the deployment is finished you can check your app app-name.heroku.com.
In this case we chose streamlit-heroku-app as our app name so if you need to navigate to streamlit-heroku-app.heroku.com to see the deployed streamlit app.

Note: the app is no longer available on heroku

Writing DAX Formulas

Kevork Keheian — Thu, 13 Aug 2020 10:43:02 +0000

A - Introduction

Every Dax formula, similar to programming variables should be defined/declared by a name and followed by the formula/value.

<Calculation name> = <DAX formula>

For example:

Order Type = "Allocation"

Here, the Order Type is the name of the DAX formula and "Allocation" is the value or Order Type.

B - Reminder

As we mentioned in the introduction that each DAX formula, whether it is a calculated column, calculated table or a measure returns a value.
In other words "Allocation" is the returned value.

This DAX formula is considered as a measure. So, whenever we drag this measure into a values field, it will return the value "Allocated" to each field of the visual.

So the returned result of a DAX will be "Allocation" for each row.

C - Referring a Table

With DAX you can duplicate a table or a column using the same method as the above, but instead of a text inside a quotation you use the name of the table.

For Example:

Let's suppose that you have a table called Sales in your model. And you want to duplicate the table using a Calculated Table you will use the below formula:

Sales Orders = Sales

Here the value or Sales Order which is the Sales table that was already in the model does not need single quotes.

Single quotes are needed when the value is a reserved word such as Date or the table name does not include embedded spaces.
EX:

Sales Orders = 'Sales Table'

Sales Dates = 'Date'

D - Referring a Column

Referred Columns in a formula should be enclosed within square brackets.

example:

Total Sales = SUM([Amount])

In the above example the value in the SUM function is the column name.
The column names are unique in the table. Some references tell you that you can use the column without preceding the table name.
But, to improve the readability of your formulas, it is preferable that the column reference precedes with its table name.

example:

Total Sales = SUM(Sales[Amount])

Another reason for preceding the table name in a column reference formula is that you never know when you will update your model. As we mentioned earlier, the column name is always unique in a table, but can be found multiple times in a model which includes multiple tables.

E - Referring a Measure:

Finally, you can refer a measure in your formula.
Let's suppose that we have the below column referring formulas (they are called measures):

Total Sales = SUM(Sales[Amount])

Total Purchases = SUM(Purchases[Amount])

And you want to calculate the difference between Sales amount and Purchase Amount.
To do this there are multiple ways:

Revenue = SUM(Sales[Amount]) - SUM(Purchase[Amount])

Revenue= [Total Sales] - [Total Purchases]

Note: This is not the right way to calculate the profit, but just an example to show how to refer to a measure in DAX

Note: Measures are a model-level object.

Although it is not recommended, but you can also precede a measure reference in a DAX formula with its table name.

example:

Revenue = [Total Sales] - Sales[Total Purchases]

As we mentioned earlier many reasons to precede the table name in column referring measure, another reason is to differentiate between column referring and measure referring formulas.

Summary:

You can refer a TABLE in a DAX Formula Sales Orders = Sales
You can refer a COLUMN in a DAX Formula Total Sales = SUM([Amount])
You can refer a MEASURE in a DAX Formula Revenue = [Total Sales] - [Total Purchases]

Introduction to DAX

Kevork Keheian — Sat, 01 Aug 2020 11:16:42 +0000

DAX is mainly used to add calculations to the Power BI Model
You can add 3 types of calculations:

Calculated Tables
Calculated Columns
Measures

It can also be used to define RLS (Row-Level Security)

1 - Calculated Tables:

Regarding the calculated tables, DAX is used to whether duplicate or transform an existing model data, create a series data or even produce a new table.
Cons: It increases the model storage size => could prolong the data refresh durations:
Examples: Date tables, Role playing dimensions, what-if analysis, etc...

Note: Calculated table can't connect to external data. You need to use Power Query to do that.

a - Date Table:

Date tables are mostly used for special filters as time intelligence.
For example if you have a model with missing dates and you want to calculate the average per date and the calculation requires the missing dates (ex: holidays), you need to create a date table table with one to many relation to do an accurate calculation.
EX: CALENDAR and CALENDARAUTO

b - Role-playing dimensions:

The Date table can also be defined as role-playing dimension.
This happens when table has two date fields that are related to Date column in Date Table.
(EX: [Sales].OrderedDate and [Sales].DeliveredDate ----> [DateTable].Date
In this case the Date table's Date column is playing the role of Ordered Date and Delivered Date of Sales Table.

Note: Power BI models allow only one active relationship between tables.

Source: Microsoft Learn

So, if you create 2 relations between 2 tables in the Power BI model one of them will be active (solid line) and the other not (dotted line).
But the only way you can use this inactive relation using USERELATIONSHIP DAX function.
Another way is to create 2 date tables.

Source: Microsoft Learn

c - What-if analysis:

What-if analysis is a featured supported in Power BI. It is a parameter that when created automatically adds a calculated table to the model.
They allow selection of filter by value stored in the calculated tables.

2 - Calculated Columns:

As we mentioned earlier, DAX can also be used to add calculated columns.

//This is not an actual DAX formula
//But it represents the concept
  foreach(row in table) {
      return value;
  }

It only differs if the connection is Import or Direct Query.
When a calculated column is added to an Import storage mode table, the calculated column is evaluated on refresh which increases the size of the model. But when added to a Direct Query storage mode the calculation is evaluated by the underlying source database upon querying the table.

3 - Measures:

Finally, you can write DAX formulas to add measures. It is used to summarize the model data.

For example: Revenue = SUM(Sales[Sales Amount])

Where Sales is the name of the table and Sales Amount is the name of the column that we are calculating.
Unlike calculated columns the returned value of a measure is never stored in the model and the formula is assessed at query time. So, the storage size of the model is affected.

Computational Thinking - Riddle 1

Kevork Keheian — Thu, 11 Jun 2020 22:05:03 +0000

What two whole, positive numbers that have a one-digit answer when multiplied and a two-digit answer when added?
riddle source: http://riddles.com/

How to?

We need to declare 2 variable sum and mult:

sum is to check the sum of the 2 digits.
mult is to check the multiplication of the 1 digit.

And both should be assigned as 0.

sum = 0
mult = 0

Now we need 2 for loops to iterate through the needed numbers
The numbers should be between 1 and 9, beacuse their multiplication should be only 1 digit.
1-Any number multiplied by 0 will result 0

2-Any number less than 10 multiplied by 10 or a number greater than 10 will result to number of 2 digits

for x in range(1, 10):
    for y in range(1, 10):
        sum = x + y
        mult = x * y
        if(sum >= 10):
            if(mult < 10):
                print(str(x) + " and " + str(y))
        break

And finally we need to break the iteration at the end of the second loop to avoid iterating through the same numbers twice.

Answer


 > 9 and 1

How to accelerate your Computational Thinking

Kevork Keheian — Thu, 11 Jun 2020 21:50:17 +0000

Computational thinking is the practice of systematically solving a problem and of constructing and presenting a solution in such a manner that a machine can execute it. Yet to think like a computer scientist, does not require an IT background.
Accelaratig your computational thinking just needs some (and sometimes a lot) of practice.
In this coming weeks I will be posting riddles, most of it mathematical riddles, and how to solve them by writing a few lines of code.