cache dataset

app.py

@@ -1,6 +1,11 @@
 import streamlit as st
 import nannyml as nml
-from sklearn.metrics import f1_score
+
+@st.cache_resource
+def get_data():
+    reference_df, analysis_df, analysis_target_df = nml.load_synthetic_car_loan_dataset()
+    return reference_df, analysis_df, analysis_target_df
+
 
 st.title('Is your model degrading?')
 st.caption('### :violet[_Estimate_] the performance of an ML model. :violet[_Without ground truth_].')
@@ -25,8 +30,7 @@ new labeled data. In this demo, we show the **Confidence-based Performance Estim
 the performance of **classification** models.
 """)
 
-reference_df, analysis_df, analysis_target_df = nml.load_synthetic_car_loan_dataset()
-test_f1_score = f1_score(reference_df['repaid'], reference_df['y_pred'])
+reference_df, analysis_df, analysis_target_df = get_data()
 
 st.markdown("#### The prediction task")