Update app.py

app.py

@@ -102,13 +102,14 @@ def _(catalog):
 
 @app.cell
 def _(pl):
-    pl.scan_csv("yellow_tripdata_2015-01.csv").group_by("passenger_count").len().sort("passenger_count").collect()
+    df_orig = pl.read_csv("yellow_tripdata_2015-01.csv.zip")
+    df_orig.group_by("passenger_count").len().sort("passenger_count").collect()
     return
 
 
 @app.cell
 def _(pl):
-    pl.read_csv("yellow_tripdata_2015-01.csv").group_by("passenger_count").len().sort("passenger_count")
+    df_orig.group_by("passenger_count").len().sort("passenger_count")
     return
 
 
@@ -120,24 +121,19 @@ def _(mo):
 
 @app.cell
 def _(pl):
-    pl.read_csv("yellow_tripdata_2015-01.csv").write_parquet("taxi.parquet", partition_by=["passenger_count"], statistics=True)
+    df_orig.write_parquet("taxi.parquet", partition_by=["passenger_count"], statistics=True)
     return
 
 
 @app.cell
 def _(pl):
-    pl.scan_parquet("taxi.parquet").group_by("passenger_count").len().sort("passenger_count").collect()
+    df_orig.group_by("passenger_count").len().sort("passenger_count").collect()
     return
 
 
 @app.cell
 def _(pl):
-    pl.read_parquet("taxi.parquet").group_by("passenger_count").len().sort("passenger_count")
-    return
-
-
-@app.cell
-def _():
+    df_orig.group_by("passenger_count").len().sort("passenger_count")
     return