Merge pull request #31 from transferwise/minor_fixes

Minor fixes: add a config json, fix bug in Bayes plotting, adjust readme

README.md

@@ -153,7 +153,9 @@ Then navigate to the folder `./tw_experimentation/streamlit`.
 
 Now run the command `streamlit run Main.py` and the app should open in your browser.
 
-Then you can use or test dataset from data/test_data.csv
+For testing, you can load the test dataset from `data/test_data.csv`. You should either set the fields as shown in the 
+screenshot below; or just click on the "Load configuration from json" section and load the whole 
+configuration from the file `data/test_config.json`.
 
 ![plot](https://github.com/transferwise/tw-experimentation/blob/main/docs/images/data_loading_streamlit.png?raw=True)
 ![plot](https://github.com/transferwise/tw-experimentation/blob/main/docs/images/evaluation_streamlit.png?raw=True)

data/test_config.json

@@ -0,0 +1 @@
+{"exp_name": "", "is_experiment": true, "variant_name": "T", "timestamp": "trigger_dates", "is_dynamic_experiment": true, "pre_experiment": ["pre_exp_revenue"], "segments": ["segment_1", "segment_2", "segment_3", "currency", "country_of_origin"], "outcomes": ["num_actions", "revenue", "conversion"], "remove_outliers": true}

tw_experimentation/bayes/bayes_test.py

@@ -368,7 +368,7 @@ def fig_posterior_cdf_by_target(self, target: str, facet_rows_variant=False):
 
         fig = self.fig_posterior_difference_cdf(
             sample_per_variant,
-            posterior_hdi_per_variant,
+            # posterior_hdi_per_variant,
             facet_rows_variant=facet_rows_variant,
         )
         fig.update_layout(