shared_utils to calitp_data_analysis for starter_kit/, don't touch tutorials yet

Author: tiffanychu90

starter_kit/gtfs_utils_v2_examples.ipynb

@@ -33,7 +33,7 @@
     "\n",
     "from siuba import *\n",
     "\n",
-    "import gtfs_utils_v2\n",
+    "from shared_utils import gtfs_utils_v2\n",
     "analysis_date = datetime.date(2023, 1, 17)"
    ]
   },

starter_kit/shared_utils_examples.ipynb

@@ -52,8 +52,8 @@
     "from calitp_data_analysis.tables import tbls\n",
     "from siuba import *\n",
     "\n",
-    "from calitp_data_analysis import geography_utils\n",
-    "from shared_utils import geography_utils, utils"
+    "from calitp_data_analysis import geography_utils, utils\n",
+    "from shared_utils import portfolio_utils"
    ]
   },
   {
@@ -424,7 +424,7 @@
     "count_cols = [\"pickup\"]\n",
     "nunique_cols = [\"pickup_zone\"]\n",
     "\n",
-    "by_borough = geography_utils.aggregate_by_geography(\n",
+    "by_borough = portfolio_utils.aggregate_by_geography(\n",
     "    df[df.pickup_borough.notna()], \n",
     "    group_cols=group_cols,\n",
     "    sum_cols = sum_cols,\n",
@@ -749,7 +749,7 @@
     }
    ],
    "source": [
-    "df2 = geography_utils.aggregate_by_geography(\n",
+    "df2 = portfolio_utils.aggregate_by_geography(\n",
     "    df[(df.payment.notna()) & (df.pickup_borough.notna())], \n",
     "    group_cols = [\"pickup_borough\", \"payment\"],\n",
     "    sum_cols = [\"passengers\", \"fare\"],\n",
@@ -981,9 +981,7 @@
    "source": [
     "import branca\n",
     "import geopandas as gpd\n",
-    "import pandas as pd\n",
-    "\n",
-    "from calitp_data_analysis import geography_utils"
+    "import pandas as pd"
    ]
   },
   {

starter_kit/simple_dask.py

@@ -1,10 +1,9 @@
 # Simple script to test dask clutser
 import dask.dataframe as dd
-import gcsfs
 import os
 import pandas as pd
 
-from calitp_data_analysis import get_fs
+from calitp_data_infra.storage import get_fs
 fs = get_fs()
 
 RT_GCS = 'gs://calitp-analytics-data/data-analyses/rt_delay/compiled_cached_views/'