Merge branch 'main' into basic_ci

Author: juanitorduz

README.md

@@ -15,8 +15,7 @@ Another distinctive feature of this package is the ability to use different mode
 
 |                           | [CausalImpact](https://google.github.io/CausalImpact/) from Google | [GeoLift](https://github.com/facebookincubator/GeoLift/) from Meta | CausalPy from [PyMC Labs](https://www.pymc-labs.io) |
 |---------------------------|--------------------------------|---------|----------------------------------------|
-| Interrupted time series   | ✅                              | ❌       | ✅                                      |
-| Synthetic control         | ❌                              | ✅       | ✅                                      |
+| Synthetic control         | ✅                              | ✅       | ✅                                      |
 | Regression discontinuity  | ❌                              | ❌       | ✅                                      |
 | Difference in differences | ❌                              | ❌       | ✅                                      |
 | Language                  | R (but see [tfcausalimpact](https://github.com/WillianFuks/tfcausalimpact))  | R       | Python                                 |
@@ -92,22 +91,6 @@ This is appropriate when you have multiple units, one of which is treated. You b
 
 > The data (treated and untreated units), pre-treatment model fit, and counterfactual (i.e. the synthetic control) are plotted (top). The causal impact is shown as a blue shaded region. The Bayesian analysis shows shaded Bayesian credible regions of the model fit and counterfactual. Also shown is the causal impact (middle) and cumulative causal impact (bottom).
 
-### Interrupted time series
-This is appropriate when you have a single treated unit, and therefore a single time series, and do _not_ have a set of untreated units.
-
-| Time | Outcome |
-|------|-----------|
-| 0    | $y_0$ |
-| 1    | $y_1$ |
-|$\ldots$ | $\ldots$  |
-| T    | $y_T$ |
-
-| Frequentist | Bayesian |
-|--|--|
-| ![](img/interrupted_time_series_skl.svg) | ![](img/interrupted_time_series_pymc.svg) |
-
-> The data (treated and untreated units), pre-treatment model fit, and counterfactual (i.e. the synthetic control) are plotted (top). The causal impact is shown as a blue shaded region. The Bayesian analysis shows shaded Bayesian credible regions of the model fit and counterfactual. Also shown is the causal impact (middle) and cumulative causal impact (bottom).
-
 ### Difference in Differences
 
 This is appropriate when you have a single pre and post intervention measurement and have a treament and a control group.

causalpy/data/simulate_data.py

@@ -67,7 +67,7 @@ def generate_synthetic_control_data(
     return df, weightings_true
 
 
-def generate_interrupted_time_series_data(
+def generate_time_series_data(
     N=100, treatment_time=70, beta_temp=-1, beta_linear=0.5, beta_intercept=3
 ):
     x = np.arange(0, 100, 1)

causalpy/pymc_experiments.py

@@ -113,6 +113,7 @@ def __init__(
         self.post_impact_cumulative = self.post_impact.cumsum(dim="obs_ind")
 
     def plot(self):
+
         """Plot the results"""
         fig, ax = plt.subplots(3, 1, sharex=True, figsize=(7, 8))
 
@@ -193,21 +194,18 @@ class SyntheticControl(TimeSeriesExperiment):
 
     expt_type = "Synthetic Control"
 
-    def plot(self):
+    def plot(self, plot_predictors=False):
         """Plot the results"""
         fig, ax = super().plot()
-        # plot control units as well
-        ax[0].plot(self.datapre.index, self.pre_X, "-", c=[0.8, 0.8, 0.8], zorder=1)
-        ax[0].plot(self.datapost.index, self.post_X, "-", c=[0.8, 0.8, 0.8], zorder=1)
+        if plot_predictors:
+            # plot control units as well
+            ax[0].plot(self.datapre.index, self.pre_X, "-", c=[0.8, 0.8, 0.8], zorder=1)
+            ax[0].plot(
+                self.datapost.index, self.post_X, "-", c=[0.8, 0.8, 0.8], zorder=1
+            )
         return (fig, ax)
 
 
-class InterruptedTimeSeries(TimeSeriesExperiment):
-    """A wrapper around the TimeSeriesExperiment class"""
-
-    expt_type = "Interrupted Time Series"
-
-
 class DifferenceInDifferences(ExperimentalDesign):
     """A class to analyse data from Difference in Difference settings.
 

causalpy/skl_experiments.py

@@ -21,7 +21,14 @@ def __init__(self, prediction_model=None, **kwargs):
 
 
 class TimeSeriesExperiment(ExperimentalDesign):
-    def __init__(self, data, treatment_time, formula, prediction_model=None, **kwargs):
+    def __init__(
+        self,
+        data,
+        treatment_time,
+        formula,
+        prediction_model=None,
+        **kwargs,
+    ):
         super().__init__(prediction_model=prediction_model, **kwargs)
         self.treatment_time = treatment_time
         # split data in to pre and post intervention
@@ -141,27 +148,18 @@ def plot_coeffs(self):
         )
 
 
-# InterruptedTimeSeries and SyntheticControl are basically the same thing but with
-# different predictor variables. So we just have a TimeSeriesExperiment class and
-# InterruptedTimeSeries and SyntheticControl are both equal to the TimeSeriesExperiment
-# class
-
-
-class InterruptedTimeSeries(TimeSeriesExperiment):
-    """A wrapper around the TimeSeriesExperiment class"""
-
-    pass
-
-
 class SyntheticControl(TimeSeriesExperiment):
     """A wrapper around the TimeSeriesExperiment class"""
 
-    def plot(self):
+    def plot(self, plot_predictors=False):
         """Plot the results"""
         fig, ax = super().plot()
-        # plot control units as well
-        ax[0].plot(self.datapre.index, self.pre_X, "-", c=[0.8, 0.8, 0.8], zorder=1)
-        ax[0].plot(self.datapost.index, self.post_X, "-", c=[0.8, 0.8, 0.8], zorder=1)
+        if plot_predictors:
+            # plot control units as well
+            ax[0].plot(self.datapre.index, self.pre_X, "-", c=[0.8, 0.8, 0.8], zorder=1)
+            ax[0].plot(
+                self.datapost.index, self.post_X, "-", c=[0.8, 0.8, 0.8], zorder=1
+            )
         return (fig, ax)
 
 

docs/examples.rst

@@ -6,16 +6,8 @@ Synthetic Control
 
    notebooks/sc_pymc.ipynb
    notebooks/sc_skl.ipynb
-
-
-Interrupted Time Series
-=======================
-
-.. toctree::
-   :titlesonly:
-
-   notebooks/its_pymc.ipynb
-   notebooks/its_skl.ipynb
+   notebooks/sc2_pymc.ipynb
+   notebooks/sc2_skl.ipynb
 
 
 Difference in Differences

docs/index.rst

@@ -74,13 +74,6 @@ This is appropriate when you have multiple units, one of which is treated. You b
 
 .. image:: ../img/synthetic_control_pymc.svg
 
-Interrupted time series
-"""""""""""""""""""""""
-
-This is appropriate when you have a single treated unit, and therefore a single time series, and do not have a set of untreated units.
-
-.. image:: ../img/interrupted_time_series_pymc.svg
-
 Difference in differences
 """""""""""""""""""""""""