Merge pull request #99 from jvivian/jvivian/issue98

Fix SCM issues (resolves #98)

dfmdash/streamlit/pages/4_Synthetic_Control_Model.py

@@ -69,6 +69,8 @@ def get_results(df, outcome_var, treatment_time, treated_unit):
     outcome_var = st.selectbox("Outcome Variable", df.columns)
     treatment_time = st.date_input("Treatment Time", value=min_time, min_value=min_time, max_value=max_time)
 
+    invert = st.checkbox("Invert Factor")
+
     submit = st.form_submit_button()
 
 if not submit:
@@ -109,6 +111,10 @@ def get_results(df, outcome_var, treatment_time, treated_unit):
 ys = synth[0, :]
 y = df[df.State == treated_unit][outcome_var]
 
+if invert:
+    y *= -1
+    ys *= -1
+
 st.subheader("Original")
 fig = go.Figure()
 fig.add_trace(go.Scatter(x=x, y=y, mode="lines", name=f"{treated_unit} {outcome_var}"))
@@ -123,20 +129,31 @@ def get_results(df, outcome_var, treatment_time, treated_unit):
 normalized_treated_outcome = normalized_treated_outcome[:, 0]
 normalized_synth = np.zeros(data.periods_all)
 most_extreme_value = np.max(np.absolute(normalized_treated_outcome))
+
+if invert:
+    normalized_treated_outcome *= -1
+    normalized_synth *= -1
 fig = go.Figure()
-fig.add_trace(go.Scatter(x=x, y=normalized_treated_outcome, mode="lines", name=f"Synthetic {outcome_var}"))
 fig.add_trace(
-    go.Scatter(x=x, y=normalized_synth, mode="lines", name=f"{treated_unit} {outcome_var}", line=dict(dash="dot"))
+    go.Scatter(
+        x=x,
+        y=normalized_synth,
+        mode="lines",
+        name=f"{treated_unit} {outcome_var}",
+    )
+)
+fig.add_trace(
+    go.Scatter(x=x, y=normalized_treated_outcome, mode="lines", name=f"Synthetic {outcome_var}", line=dict(dash="dot"))
 )
 fig.add_vline(treatment_time, name=f"{outcome_var} Response")
 st.plotly_chart(fig, use_container_width=True)
 
-st.subheader("Cumulative")
-cumulative_effect = np.cumsum(normalized_treated_outcome[data.periods_pre_treatment :])
-cummulative_treated_outcome = np.concatenate((np.zeros(data.periods_pre_treatment), cumulative_effect), axis=None)
+# st.subheader("Cumulative")
+# cumulative_effect = np.cumsum(normalized_treated_outcome[data.periods_pre_treatment :])
+# cummulative_treated_outcome = np.concatenate((np.zeros(data.periods_pre_treatment), cumulative_effect), axis=None)
 
-fig = go.Figure()
-fig.add_trace(go.Scatter(x=x, y=normalized_synth, mode="lines", name=f"Synthetic {outcome_var}", line=dict(dash="dot")))
-fig.add_trace(go.Scatter(x=x, y=cummulative_treated_outcome, mode="lines", name=f"{treated_unit} {outcome_var}"))
-fig.add_vline(treatment_time, name=f"{outcome_var} Response")
-st.plotly_chart(fig, use_container_width=True)
+# fig = go.Figure()
+# fig.add_trace(go.Scatter(x=x, y=normalized_synth, mode="lines", name=f"Synthetic {outcome_var}", line=dict(dash="dot")))
+# fig.add_trace(go.Scatter(x=x, y=cummulative_treated_outcome, mode="lines", name=f"{treated_unit} {outcome_var}"))
+# fig.add_vline(treatment_time, name=f"{outcome_var} Response")
+# st.plotly_chart(fig, use_container_width=True)