Cohort Analysis with Python from Scratch | Easy Code

[@casmiranyaegbu9945]

This tutoria was quite grannular. I now have a better understanding of Cohort Analysis. Thank you

[@absentdata]

Thank you for watching

[@alexkim7270]

Dear good sir I loved this pace so much. It is much easier to understand everything from top to bottom than asking ChatGPT to spit the code out (was rushing for time hence not much time to digest everything given).

[@absentdata]

Glad to hear that

[@jessicahaire4024]

Awesome code along session -- pace was great! Thanks for explaining all the steps and subtleties, and for pointing us to the dataset on the UCI repository.

[@absentdata]

You're very welcome! Yes, I tried to keep the pace slow.

[@ju-lyndav7087]

Thank you so much. This has really helped me understand things.

[@samfreeman8328]

Thanks a lot for your heat map tutorial!

[@absentdata]

Thanks for watching!

[@SanthoshKumar-kx1xh]

Beautifully explained.

[@absentdata]

Thanks for the great compliment. Share the vid if you think if will help someone

[@NamPham15019]

Amazing video, I am a fan of your contents. Would it be possible if you can have some of the intermediate and advanced videos about market segments, revenue and cohort, and predictive analysis please. Truly appreciate it

[@youyou8803]

Amazing channel, would love to watch more videos like this. :)

[@absentdata]

More to come!

[@ronenTheBarbarian]

Thanks a lot for this! Really simple code and great explanations throughout. Keep up the great content fellow data person.

[@saya5664]

Great video! thank you and looking forward to more videos like this too :)

[@absentdata]

Thank you! Will do!

[@leonardofreitas8900]

Could you please provide a more detailed explanation of the outlier displayed at the end. Could it be caused by a problem with the data? Thank you for the video! Great content!

[@followrajah7743]

Thank you!

[@ahmadel-ashery8860]

Amazing Thank you

[@absentdata]

Your Welcome 😊

[@djaadiabdellah9081]

Great video

[@absentdata]

Glad you liked it

[@FindingyourKey]

Thanks so much for this great video. I had a question about Invoice date. In my use case, We have a purchase date of when the client bought our monthly subscription and cancel date for when they cancelled it. Can I swap Invoice month to cancellation month and use the same method?

[@absentdata]

If I am understanding the scenario correctly. Then yes you can.

[@ohreinaldo1990]

Thanks for this awesome video and explanation! Just a question... would the same rational work for a dataframe with a single customer ID per month (1 customerID per row, per month) and two columns (activation date / exit date)? Pls take into consideration that null exit date means the customer is still active... The idea is basically the same, which is check for how long my customers are active. Thank you in advance!

[@absentdata]

Hi thanks for the comment. Yes you could do that. However, evaluating how a customer is still active would only require grouping customer with a null exit date and comparing the start date to the current date.

[@akwamfoneventus6124]

Thank you for this tutorial, this has been most helpful.
But I have some things I would love to change
- How do I make the values in 100s show as whole numbers rather than the "1.2e+02" format ?
- How do I move the cohort index to the top of the chart instead of te botom on the visuals
- How can I make the plot interactive such that it will show the customerid that made up a certain cell in the cohorts ?
Thank you

[@akwamfoneventus6124]

You can get rid of the scientific notation in the heatmap specifying fmt='.2f' or fmt='.0f'

[@AffairsBibes]

Thanks ❤❤

[@mohamed.montaser]

why you didn't transform cohort_data to dataframe?

[@giandenorte]

First!

[@absentdata]

Second! :)