Data Analysis with Python — and Geospatial Data Analysis & Visualization

A complete, eleven-class course that takes an absolute beginner from print('Hello World') to a publication-ready geospatial analysis — trend maps, spatial statistics and journal-quality figures at 300 dpi.

Every notebook runs end to end on the data in this repository. Nothing has to be downloaded, and nothing has to be invented.

Keynote Speaker: Md Wabidur Rahman, Researcher, Research and Integrated Thought (RIT), Dhaka, Bangladesh

A Bengali edition of this course is at data-analysis-with-python-bn. The code is identical; only the explanations differ.

---

The classes

Part 1 — Python for data analysis (Classes 01–07)

#	Notebook	What it covers
01	Class_01_Introduction_Variables_DataTypes_Operators_001.ipynb	Variables, data types, arithmetic and assignment operators, casting, f-strings, the three errors beginners meet
02	Class_02_Operators_Comparison_Logical_Identity_Membership_Bitwise.ipynb	Comparison, logical, identity, membership and bitwise operators; why & and not and in pandas
03	Class_03_Lists_Tuples_Sets_Dictionaries.ipynb	The four containers, copy() versus =, zip(), applied to a real rainfall series
04	Class_04_Booleans_Conditionals_Loops.ipynb	Booleans, if/elif/else, nested conditions, for, while, break, continue
05	Class_05_Functions_Lambda_Strings_Classes_Modules.ipynb	Functions, *args, defaults, recursion, lambda, string methods, classes, modules
06	Class_06_NumPy_and_Pandas.ipynb	ndarrays, reshaping, vectorised arithmetic, boolean indexing; Series, DataFrame, loc/iloc, reading Excel
07	Class_07_Missing_Data_and_Statistical_Analysis.ipynb	Sentinel values, isnull, dropna, fillna, interpolate, describe, corr, resample

Part 2 — Geospatial analysis (Classes 08–11)

These four notebooks form one continuous study, answering a single research question:

How did near-surface air temperature and relative humidity change over Bangladesh between 2000 and 2024, and what is the spatial structure of that change?

#	Notebook	What it covers
08	Class_08_Foundations_of_Spatial_Data.ipynb	Vector versus raster, Shapefile/GeoJSON/GeoPackage/GeoParquet benchmarked, shapely geometry, NetCDF and ERA5, a first map
09	Class_09_Vector_Data_Manipulation.ipynb	CRS and the area disaster, centroids, spatial joins, the coarse-grid problem, dissolve/overlay/clip/buffer, geocoding
10	Class_10_Raster_and_Remote_Sensing.ipynb	rioxarray, affine transforms, reprojection, clipping with all_touched, zonal statistics, GeoTIFF and COG, STAC — satellite imagery without downloading, NDVI
11	Class_11_Advanced_Analysis_and_Visualization.ipynb	Area-weighted trends, vectorised per-pixel OLS, Mann-Kendall and Sen's slope, Moran's I, LISA, Getis-Ord Gi*, K-means climate zones, networks, folium, journal-ready Figures 1–3 and Tables 1–3

By the end of Class 11 the outputs/ folder holds three figures at 300 dpi in PNG and PDF, three CSV tables, a draft Results paragraph and a reproducibility record — the Results section of a paper.

---

Getting started

1. Create the environment

The geospatial libraries sit on top of GDAL, GEOS and PROJ, which are C/C++ libraries. Installing them all from conda-forge keeps their versions matched; mixing channels, or installing them with pip on Windows, is the usual cause of a broken setup.

conda env create -f environment.yml
conda activate geo

Or build it by hand:

conda create -n geo -c conda-forge python=3.12 numpy pandas xarray rioxarray rasterio \
  geopandas pyogrio shapely pyproj netcdf4 h5netcdf matplotlib scipy scikit-learn dask \
  libpysal esda splot mapclassify networkx folium contextily pyarrow geopy \
  jupyterlab notebook ipykernel -y

conda activate geo
python -m pip install pymannkendall pystac-client odc-stac planetary-computer

2. Start Jupyter from inside the environment

conda activate geo
jupyter lab

On Windows this matters more than it sounds. If you launch Jupyter from base and then switch the kernel to geo, the kernel inherits base's PATH and can load the wrong GDAL/PROJ DLLs — which kills the kernel with no error message. Launch Jupyter from the same environment the notebooks will run in.

3. Run

Open any notebook and use Kernel → Restart Kernel and Run All Cells. The geospatial notebooks find and unpack the data archives by themselves on first run.

---

The data

File	What it is	Source	Licence
monthly_mean_data_5.3.25.zip	ERA5 monthly means, temperature and relative humidity at 1000 hPa, 300 months (2000–2024), 0.25° over 20–27°N / 87–94°E	Copernicus Climate Data Store	Copernicus Licence — free reuse with attribution
hourly_mean_data_5.3.25.zip	ERA5 monthly means by hour of day (the diurnal cycle)	Copernicus CDS	Copernicus Licence
bgd_adm_bbs_20201113_SHP.zip	Official administrative boundaries of Bangladesh (BBS, 2020): 8 divisions, 64 districts, 544 upazilas	HDX / Bangladesh Bureau of Statistics	see the dataset page on HDX
Data.xlsx, Shundorbon.xlsx, Khulna.csv	Daily weather series derived from NASA POWER	NASA POWER	freely available
sample1.csv, barishal_temperature.txt, bangladesh_seawater.txt	Small teaching datasets	prepared for this course	MIT, with the code
Boolean_Algebra.pdf	Reference handout for Class 02	prepared for this course	CC BY 4.0

Getting the GADM file

Class 08 compares two boundary datasets — GADM 2015 (7 divisions, 65 districts) against BBS 2020 (8 divisions, 64 districts) — to show why the source and version of a boundary file must be stated in a paper's Methods.

GADM is not included here, because its licence says "Redistribution or commercial use is not allowed without prior permission." To run that one comparison, download the Bangladesh shapefile yourself from gadm.org and place it next to the notebooks as BGD_adm.zip, with layers named BGD_adm0.shp … BGD_adm3.shp.

Without it, Class 08 prints a short notice, shows the BBS boundaries alone, and carries on. Nothing else in the course depends on GADM.

---

What is in the repository

.
├── Class_01 … Class_11 .ipynb     the eleven notebooks
├── *.zip, *.xlsx, *.csv, *.txt    the data, read directly by the notebooks
├── environment.yml                the conda environment
├── requirements.txt               a pip fallback
├── LICENSE
└── outputs/                       created on the first run (git-ignored)

Everything sits at the top level on purpose: the notebooks read their data by bare filename (pd.read_csv('sample1.csv')), so a data/ subfolder would break them.

The notebooks are committed without stored outputs, which keeps the diffs readable and the repository small. Run them once and every figure and table appears.

---

Licence

The notebooks, the code and the teaching text are released under the MIT Licence (see LICENSE). The datasets keep their own licences, listed in the table above — GADM in particular is not redistributed here.

If this material is useful in your teaching or research, a citation is appreciated:

Rahman, M. W. (2026). Data Analysis with Python, and Geospatial Data Analysis & Visualization: an eleven-class course. Research and Integrated Thought (RIT), Dhaka.

---

Acknowledgements

ERA5 data were generated using Copernicus Climate Change Service information. Neither the European Commission nor ECMWF is responsible for any use of the Copernicus information contained here. Administrative boundaries come from the Bangladesh Bureau of Statistics via the Humanitarian Data Exchange.