Integrating Python in Power BI With An Example Of Data Wrangling & Data Visualization

-

 


With a mix of Power BI & Excel's versatility and coding experience, You can easily wrangle any kind of data and get the most out of it.

Let's see how we can use python in two processes (Data Wrangling & Data Visualization) inside Power BI

  • Make sure that anaconda and python3 installed on your pc.


  1. Our Data set

The dataset I'm using for this article is a Kaggle dataset that I've recently worked on you can check it out here
what we are going to do is to transform some of my work in this notebook to power bi.


  1. 2. Data wrangling

The data wrangling process in power bi is based inside the query editor

After loading your data click on the transform tab then run a python script
then you will need to import your libraries and run your code
for me, the dataset only needed to replace some values and to get the date and time 
for more information check out my Kaggle notebook

You will get this after running your code
this is the new dataset after running your code 
so all you have to do is to remove the named column as we don't need it 
and expand the value column which holds our data

Here's our clean data ready to be loaded

  1. 3. Data Visualization 
Now let's get to the fun part, From my notebook I've selected two visuals that I wanted to show in my power bi file
1- select the python visual
2- load your data into it
3- write your code
4- run

 But here are some important notes you should know:
  • Don't forget to import your libraries (numpy as np, pandas as pd, matplotlib.pyplot as plt, seaborn as sb....etc)
  • Your dataset has its name now as 'dataset', not 'df' or else. you will have to change it if you want or keep it as it is.
  • Don't forget plt.show() at the end this is our magic word

Here's another one


Now you are good to go.
Regards.

Comments

Post a Comment

Popular posts from this blog

Get The Distinct Count Of Values By Each Day Using DAX

-
Image
  We have a serious issue when it comes to calculating the distinct count of values based on dates, why? cause if you want to get the distinct count for each day, the usual distinct count measure will ignore the day-by-day calculating and will get the distinct count for the whole period.  so let's see how to overcome this. The solution is simple ... we will create a summarization by date and we will iterate through it calculating the distinct count This is the original measure that calculates the distinct count This is our new adjusted measure it simply defines a variable that holds a summarization of our table by date  and through this summarization , we will iterate and calculate the sum of the distinct count for each day Here we can see that for one day the two measures give the same value and it's reasonable  This is the tricky part that the new measure overcomes, for more days, the old measure gets the unique values for the whole bunch of days, not for each day....
Read more

Tips To Overcome Over plotting of Dense Scatter Plots.

-
Image
Here are some tricks to overcome overplotting of dense scatter plots Use a transparency factor: # Matplotlib scatter plot with an alpha value plt . scatter ( df [ X_COL ], df [ Y_COL ], alpha = 0.03 ) plt . xlabel ( X_COL ) plt . ylabel ( Y_COL ) plt . show () This will introduce some transparency to your visual and the less transparency the denser. You can make this even cooler with animated transparency: # Initialize plot and animation camera fig , ( ax1 , ax2 ) = plt . subplots ( 1 , 2 ) # Create 2 camera = Camera ( fig ) # Create a sequence of alpha values alpha_range = np . linspace ( 0.5 , 0 , 30 ) ** 3 # For each alpha value for alpha_value in alpha_range : # Plot "vanilia" plot for reference ax1 . scatter ( df [ X_COL ], df [ Y_COL ], color = 'black' ) # Plot scatter plot with the alpha value ax2 . scatter ( df [ X_COL ], df [ Y_COL ], alpha = alpha_value , color = 'black' ) # Take a ...
Read more