00:20 Introduction to causal inference theory and practice 02:35 Common causes can explain apparent correlations. 06:58 Traditional data science pipeline vs causal pipeline 09:08 Represent problem as a causal DAG 13:30 Understanding causal effects through interventions and surgeries on the DAG arrows 15:41 Identification and back door path 19:44 Causal methods resolve paradoxes in identifying effects 21:36 Causal inference methods for regression analysis 25:25 Using a causal pipeline to analyze the effect of subscribing on spending. 27:12 Representing causal diagrams and data frame creation for causal modeling 30:45 Adjust variables to run estimation and implement estimation methods. 32:40 Control only for variables in the minimal set for computational efficiency. 36:29 Testing consistency of DAG with data using conditional independencies 38:31 Testing independence between X and Y based on partial correlation coefficient 42:22 Identifying minimal sufficient adjustment sets for estimating total effects Crafted by Merlin AI.
Without the data, I am unable to see any estimands. For all the 3 estimands it shows no such variables found. How can I know what variables to adjust ?
Good challenge - you can try to create synthetic data (column names provided) based on your assumptions for distributions/rules and see what will happen ;)
On slide 24, you mentioned that conditional on Z, if there is a significant dependence between X and Y, then the DAG is possibly wrong. I am confused why? Why could it not mean that there is actually a legit causal relationship between X and Y ?
Hey! You can access all relevant resources as referred to in this talk via our website: datasciencefestival.com/session/causal-inference-in-python-theory-to-practice/
00:20 Introduction to causal inference theory and practice
02:35 Common causes can explain apparent correlations.
06:58 Traditional data science pipeline vs causal pipeline
09:08 Represent problem as a causal DAG
13:30 Understanding causal effects through interventions and surgeries on the DAG arrows
15:41 Identification and back door path
19:44 Causal methods resolve paradoxes in identifying effects
21:36 Causal inference methods for regression analysis
25:25 Using a causal pipeline to analyze the effect of subscribing on spending.
27:12 Representing causal diagrams and data frame creation for causal modeling
30:45 Adjust variables to run estimation and implement estimation methods.
32:40 Control only for variables in the minimal set for computational efficiency.
36:29 Testing consistency of DAG with data using conditional independencies
38:31 Testing independence between X and Y based on partial correlation coefficient
42:22 Identifying minimal sufficient adjustment sets for estimating total effects
Crafted by Merlin AI.
This is a highly informative and useful presentation. It is clear, concise, and to the point.
Glad to hear it! 🎉
Where can I find the 'full_data' csv file? The drive link only contains the ipynb file and the pdf.
Without the data, I am unable to see any estimands. For all the 3 estimands it shows no such variables found. How can I know what variables to adjust ?
Good challenge - you can try to create synthetic data (column names provided) based on your assumptions for distributions/rules and see what will happen ;)
How can I download the Jupyter Notebook presented in the video?
On slide 24, you mentioned that conditional on Z, if there is a significant dependence between X and Y, then the DAG is possibly wrong. I am confused why? Why could it not mean that there is actually a legit causal relationship between X and Y ?
Where can we download the full_data.csv?
Hey! You can access all relevant resources as referred to in this talk via our website: datasciencefestival.com/session/causal-inference-in-python-theory-to-practice/
bookmarking 23:08