Python Project - Optimize Marketing Campaigns | Regression and Correlation Analysis
Вставка
- Опубліковано 20 сер 2022
- Learn how to discover which marketing tactics have the most significant effect on your sales with this easy to apply Python project. You will learn how to develop a linear regression model and identify which variables are significant.
Find the Collab notebook here:
drive.google.com/file/d/1U493...
Find the dataset here:
www.absentdata.com/wp-content...
#marketing
#python
#dataanalysis
#datascience
Nice intro project man! This was my first linear regression problem solved in python. Up until now, I've been doing everything in excel. I appreciate the video!
Congratulations on building your first linear model in Python! Share the video with anyone you think it will help.
Thank you so much for the video! Incredibly useful, well-made and not boring at all.
Great feedback. Im glad it was useful
This is a great video!
I have some follow up questions
1. Is there a way to find out the interwine effect of multiple campaigns? How do I prove that multiple campaigns will drive better results than one alone? For example, phone + email is better than phone alone.
2. How do I make budget allocation recommendations based off of this analysis? For example, what's the ideal percentage split for each campaign type?
You can use the old formula and multiply the channels this produce additional channels interactions that you can receive a coefficient for. Like email and phone in addition to just email and phone. 2. This an optimization question to get the ideal split of budget for revenue, you feed the model coefficients into the model and it will provide you the best option. Like Excel Solver
Very much appreciated
From what source did you get this dataset?
Nice presentation
Thank you
nice demonstration! good work! but I was wondering what if it was not a linear function? for example, what if you had a xgboost regression instead of a linear regression? how would you formulate your objective function?
Xboogst is an ensemble model so you can also get the coefficients from that model. So you should be able to substitute it in for linear regression
really great way of how we think
Glad you like it. Definitely share it if you think it helps someone
@@absentdata yes, sometimes we only focus on technical standpoints. but this approach will helps for user think about analytical standpoint with business, really pracical!!
cool cool cool
Very much appreciated. Can you please state the problem more clearly?
There is a 1:30 power point introduction to why this is being done. Plus at 1:54 the objective is written out and described. I think you might have missed this
@@absentdata Thank you. One more question: What is the fifth type of sales contact: I think you listed 4 of them in the presentation slide, but the data has Sales Contact 5 as well.
Oh never mind - it is the Organic Contact - I had too much wine at Thanksgiving! Happy Holidays!
Very badly explained. Couldn't understand the meaning of some variables because there is no explanation on them.
Sorry to hear that. Happy to try to answer your questions here.
@@absentdata What does the independent features/columns mean exactly i.e. columns starting with 'Campaign ()' and 'Sales Contact1,2,3...'? Is it the amount spent? What exactly is it?
@@ahsanshafiqchaudhry I have the same doubt
Also, there are Medium Facilities with more clients than Large Facilities. Why is that? What does the Number of Clients variable mean?