Code and environment cleanup, cached data loading and dynamic geojson…

… parsing

.flake8

@@ -17,5 +17,4 @@ ignore =
     ; line break after binary operator
     W504
 exclude =
-    src/archive/*
-    ; src/analysis.py
+    src/archive/*

.mypy.ini

@@ -17,18 +17,12 @@ warn_unused_configs = True
 disallow_any_generics = True
 warn_return_any = True
 warn_unreachable = True
-; Pydantic integration
-; https://docs.pydantic.dev/latest/integrations/mypy/
-; plugins =
-;   pydantic.mypy
 
-; Do not type-check the specified subpkgs/modules
 ; Inline silencing of specific errors:
 ; https://mypy.readthedocs.io/en/stable/error_codes.html
 
 ; Exclude specific files or directories
 exclude = (?x)(
-    ; analysis.py
-    ; | archive/.*
     archive/.*
+    ; | colors.py
     )

_quarto.yml

@@ -15,12 +15,9 @@ website:
   search: false
   navbar:
     title: false
-  #   background: "#eee"
     left:
       - href: notebooks/report.ipynb
         text: "Report"
-      # - href: notebooks/sql_qs.ipynb
-      #   text: "SQL exercises"
   page-footer:
     center:
       - href: https://bainmatt.github.io/
@@ -29,7 +26,3 @@ website:
 format:
   html:
     theme: ember
-    # css: styles.css
-    # theme:
-    #   light: flatly
-    #   dark: darkly

environment.yml

@@ -4,18 +4,19 @@
 #   (2.0) $ conda activate {name}
 #
 # This env was pointed to, created and activated on, an external drive by:
-# 
+#
 #   (0.0) $ conda config --append envs_dirs /Volumes/Samsung_T5/conda-environments
 #   (1.0) $ conda env create --prefix /Volumes/Samsung_T5/conda-environments/streamlit-geomaps --file environment.yml
 #   (2.1) $ conda deactivate
 #   (2.2) $ conda activate /Volumes/Samsung_T5/conda-environments/streamlit-geomaps
 #   (2.3) <setting the vscode interpreter to that of the same environment>
 #   (3.0) <setting '.env' `PYTHONPATH` variable to .:/Volumes/Samsung_T5/conda-environments/streamlit-geomaps/lib/python3.11/site-packages
-# 
+#
 # To check that the environment has been successfully activated (has *):
-# 
-#   (0.0) $ conda info --envs
-# 
+#
+#   (0.1) $ conda config --show envs_dirs
+#   (0.2) $ conda info --envs
+#
 
 name: streamlit-geomaps
 prefix: /Volumes/Samsung_T5/conda-environments/streamlit-geomaps
@@ -29,59 +30,25 @@ dependencies:
   - pip>=19.0
   - jupyter
   - pandas
-  - numpy
+  - requests
+  - python-dotenv
+  # - gdown
 
   # Dataviz
   - matplotlib
-  - seaborn
   - plotly
   - geopandas
-  - geopy
+  # - geopy
   - geojson
+  - seaborn
 
   # Stats
-  # - scikit-learn
-  # - scipy
-  # - statsmodels
-  # - arviz
-
-  # ETL
-  - requests
-  - gdown
-  # - pydantic
-  # - bs4
-  # - mysql
-
-  # Ensures pycurl and wptools are installed correctly
-  # - openssl
-  # - pycurl
-
-  # Secure API key retrieval
-  - python-dotenv
+  - scipy
 
   - pip:
-    # Stats
-    # - bambi
-
-    # Tables
-    # - great-tables
-    # - polars
-    # - pyarrow
-    # - tabulate
-
     # Dashboards
     - streamlit
 
-    # EDA
-    # - holidays
-
-    # SQL
-    # - mysql
-    # - mysqlclient
-    # - mysql-connector-python
-    # - PyMySQL
-    # - jsonschema
-
     # Typing / testing / linting
     # run `mypy --install-types` to install missing stubs. Pandas given below.
     - mypy

notebooks/report.ipynb

@@ -71,7 +71,6 @@
     "#| code-fold: true\n",
     "\n",
     "import pandas as pd\n",
-    "import matplotlib.pyplot as plt\n",
     "from IPython.display import display as display3\n",
     "\n",
     "from src.paths import get_path_to\n",
@@ -81,24 +80,18 @@
   },
   {
    "cell_type": "code",
-   "execution_count": 3,
+   "execution_count": null,
    "metadata": {},
    "outputs": [],
    "source": [
     "#| include: false\n",
     "\n",
     "customize_plots()\n",
     "%config InlineBackend.figure_format = 'svg'\n",
-    "# plt.style.use(\"_matplotlibrc\")\n",
     "\n",
     "import plotly.io as pio\n",
     "pio.renderers.default = \"notebook\"\n",
-    "# pio.renderers.default = \"plotly_mimetype+notebook_connected\"\n",
-    "\n",
-    "# pd.set_option('display.expand_frame_repr', False)\n",
-    "# pd.set_option('display.max_columns', 8, 'display.max_colwidth', 30)\n",
-    "# pd.reset_option('display.max_columns')\n",
-    "# pd.reset_option('display.max_colwidth')"
+    "# pio.renderers.default = \"plotly_mimetype+notebook_connected\""
    ]
   },
   {
@@ -120,11 +113,6 @@
     "#| code-summary: Load\n",
     "\n",
     "if 'data' not in locals():\n",
-    "    # data = pd.read_csv(\n",
-    "    #     get_path_to(\"data\", \"raw\", \"\"),\n",
-    "    #     encoding='ISO-8859-1',\n",
-    "    #     low_memory=False\n",
-    "    # )\n",
     "    pass\n",
     "else:\n",
     "    print(\"data loaded.\")"

pytest.ini

@@ -3,7 +3,6 @@
 [pytest]
 pythonpath = src
 ; minversion = "6.0"
-; addopts = -q
 addopts = -ra -vv --ignore=src/app.py --ignore-glob=*/archive/*
 testpaths =
     tests

requirements.txt

@@ -1,28 +1,22 @@
 jupyter>=1.1.1
 requests>=2.32.3
 python-dotenv>=1.0.1
+pandas>=2.2.3
 
-numpy>=1.24.2
-scipy>=1.9.3
-pandas>=2.0.0
-
-matplotlib>=3.6.2
-seaborn>=0.12.1
+matplotlib>=3.9.3
 plotly>=5.24.1
-geopandas>=0.12.1
-geopy>=2.4.1
+geopandas>=1.0.1
+# geopy>=2.4.1
 geojson>=3.1.0
+seaborn>=0.13.2
 streamlit>=1.40.2
-gdown>=5.2.0
-
-# great-tables>=0.11.1
-# polars>=0.14.28
-# pyarrow>=17.0.0
+scipy>=1.14.1
+# gdown>=5.2.0
 
 mypy>=1.13.0
 typing_extensions>=4.12.2
 pandas-stubs>=2.2.3.241126
 types-requests>=2.32.0.20241016
 pytest>=8.3.4
-coverage>=7.6.8
+coverage>=7.6.9
 flake8>=7.1.1