# Phase 1 โ€” Scientific Python Stack + Plotting (โ‰ˆ3โ€“4 weeks) **Goal:** fluency in **NumPy**, **Pandas**, **Matplotlib/Seaborn** โ€” the tools every later phase sits on. **Environment:** mostly **๐Ÿ–ฅ๏ธ Local** in Jupyter; switch to **๐Ÿ“Š Kaggle** for the EDA capstone (datasets are one click away). Each checkpoint is tagged. **๐Ÿ”„ Freshness:** libraries move โ€” use **current NumPy 2.x / pandas 2.x / Matplotlib** (links point to "stable" docs). Core APIs are steady, but if a method is deprecated, the current docs will tell you. **Primary resources** (open the linked playlist/site, then the **bold** item): - [Corey Schafer โ€” *Pandas* playlist](https://www.youtube.com/playlist?list=PL-osiE80TeTsWmV9i9c58mdDCSskIFdDS) ยท [Corey Schafer โ€” *Matplotlib* playlist](https://www.youtube.com/playlist?list=PL-osiE80TeTvipOqomVEeZ1HRrcEvtZB_) - [*Python Data Science Handbook* (free, online)](https://jakevdp.github.io/PythonDataScienceHandbook/) ยท [NumPy docs (stable)](https://numpy.org/doc/stable/) ยท [pandas docs (stable)](https://pandas.pydata.org/docs/) - [Kaggle Learn](https://www.kaggle.com/learn) (hands-on, free certs) ยท ๐Ÿ†• [Vizuara](https://www.youtube.com/@vizuara) (good "NumPy/Pandas from scratch" explainers) - ๐ŸŽจ **[Illustrated explainers](https://learn-by-visuallization.org/illustrated/index.html)** (interactive, offline) for this phase: [Vectorization & ufuncs](https://learn-by-visuallization.org/illustrated/1-scientific-stack/numpy-vectorization.html), [Broadcasting](https://learn-by-visuallization.org/illustrated/1-scientific-stack/numpy-broadcasting.html), [Aggregations & axis](https://learn-by-visuallization.org/illustrated/1-scientific-stack/numpy-axis.html), [Indexing: slices/masks/fancy](https://learn-by-visuallization.org/illustrated/1-scientific-stack/numpy-indexing.html), [Series/DataFrame & loc/iloc](https://learn-by-visuallization.org/illustrated/1-scientific-stack/pandas-dataframe.html), [GroupBy](https://learn-by-visuallization.org/illustrated/1-scientific-stack/pandas-groupby.html), [merge & join](https://learn-by-visuallization.org/illustrated/1-scientific-stack/pandas-merge.html). Pair with the visual [Visual NumPy (Alammar)](https://jalammar.github.io/visual-numpy/). --- ## Module 1A โ€” NumPy | # | Topic | Open exactly this | ~Time | |---|-------|-------------------|-------| | 1 | Arrays, dtypes, creation | [PDS Handbook](https://jakevdp.github.io/PythonDataScienceHandbook/) โ€” **Ch.2 "Understanding Data Types in NumPy"** + **"The Basics of NumPy Arrays"** ยท ๐Ÿ”— [Visual NumPy (Alammar)](https://jalammar.github.io/visual-numpy/) | 40m | | 2 | Indexing, slicing, boolean & fancy indexing | [PDS Handbook](https://jakevdp.github.io/PythonDataScienceHandbook/) โ€” **"Comparisons, Masks, and Boolean Logic"** + **"Fancy Indexing"** ยท ๐ŸŽจ [Visualize](https://learn-by-visuallization.org/illustrated/1-scientific-stack/numpy-indexing.html) | 40m | | 3 | Vectorized math (ufuncs) โ€” no loops | [PDS Handbook](https://jakevdp.github.io/PythonDataScienceHandbook/) โ€” **"Computation on Arrays: Universal Functions"** ยท ๐ŸŽจ [Visualize: vectorization](https://learn-by-visuallization.org/illustrated/1-scientific-stack/numpy-vectorization.html) ยท ๐Ÿ”— [Visual NumPy](https://jalammar.github.io/visual-numpy/) | 25m | | 4 | **Aggregations & `axis`** | [PDS Handbook](https://jakevdp.github.io/PythonDataScienceHandbook/) โ€” **"Aggregations: Min, Max, and Everything in Between"** ยท ๐ŸŽจ [Visualize](https://learn-by-visuallization.org/illustrated/1-scientific-stack/numpy-axis.html) | 25m | | 5 | **Broadcasting** (critical for ML) | [PDS Handbook](https://jakevdp.github.io/PythonDataScienceHandbook/) โ€” **"Computation on Arrays: Broadcasting"** ยท ๐ŸŽจ [Visualize](https://learn-by-visuallization.org/illustrated/1-scientific-stack/numpy-broadcasting.html) ยท ๐Ÿ”— [Visual NumPy](https://jalammar.github.io/visual-numpy/) | 30m | | 6 | Linear algebra ops: `@`, `dot`, `reshape`, `linalg` | [NumPy docs (stable)](https://numpy.org/doc/stable/reference/routines.linalg.html) โ€” **"Linear algebra (numpy.linalg)"** | 30m | **โœ… Checkpoint 1A** โ€” ๐Ÿ–ฅ๏ธ Local (notebook), no Python loops โ€” one per topic: - **(T1)** Create a `(3,4)` zeros array, an `arange` array, and a random-float array; print `.shape`/`.dtype`. - **(T2)** From `arange(10)` select evens with a **boolean mask** and `[1,3,5]` with **fancy indexing**. - **(T3)** Apply `np.exp`/`np.sqrt` to an array (a **ufunc**); time it vs a list comprehension. - **(T4)** Recreate: `np.sum([[1,2],[3,4]], axis=1)` and `axis=0`; explain each. - **(T5)** Build a 5ร—5 multiplication table by **broadcasting** a column ร— a row. - **(T6)** Multiply `(2,3)@(3,2)`; take `np.linalg.inv` of a square matrix and `np.linalg.norm` of a vector. - *(integrative)* Standardize a `(100,3)` matrix to zero-mean/unit-var **per column** using `axis`. --- ## Module 1B โ€” Pandas | # | Topic | Open exactly this | ~Time | |---|-------|-------------------|-------| | 7 | **Series & DataFrame**, loading CSVs | [Corey Schafer Pandas](https://www.youtube.com/playlist?list=PL-osiE80TeTsWmV9i9c58mdDCSskIFdDS) โ€” **Part 1** + **Part 2** ยท ๐ŸŽจ [Visualize](https://learn-by-visuallization.org/illustrated/1-scientific-stack/pandas-dataframe.html) ยท ๐Ÿ“– [PDS Handbook: Pandas Objects](https://jakevdp.github.io/PythonDataScienceHandbook/03.01-introducing-pandas-objects.html) | 60m | | 8 | `loc` vs `iloc`, indexes | [Corey Schafer Pandas](https://www.youtube.com/playlist?list=PL-osiE80TeTsWmV9i9c58mdDCSskIFdDS) โ€” **Part 2** + **Part 3 (Indexes)** ยท ๐ŸŽจ [Visualize: loc/iloc](https://learn-by-visuallization.org/illustrated/1-scientific-stack/pandas-dataframe.html) ยท ๐Ÿ“– [PDS Handbook: Data Indexing & Selection](https://jakevdp.github.io/PythonDataScienceHandbook/03.02-data-indexing-and-selection.html) | 45m | | 9 | Filtering with conditionals | [Corey Schafer Pandas](https://www.youtube.com/playlist?list=PL-osiE80TeTsWmV9i9c58mdDCSskIFdDS) โ€” **Part 4 (Filtering)** ยท ๐ŸŽจ [Visualize: boolean masks](https://learn-by-visuallization.org/illustrated/1-scientific-stack/numpy-indexing.html) ยท ๐Ÿ“– [pandas docs: Boolean indexing](https://pandas.pydata.org/docs/user_guide/indexing.html#boolean-indexing) | 25m | | 10 | Add/update/remove columns | [Corey Schafer Pandas](https://www.youtube.com/playlist?list=PL-osiE80TeTsWmV9i9c58mdDCSskIFdDS) โ€” **Part 5 & 6** ยท ๐Ÿ“– [PDS Handbook: Operations in Pandas](https://jakevdp.github.io/PythonDataScienceHandbook/03.03-operations-in-pandas.html) | 40m | | 11 | **GroupBy + aggregation** | [Corey Schafer Pandas](https://www.youtube.com/playlist?list=PL-osiE80TeTsWmV9i9c58mdDCSskIFdDS) โ€” **Part 8 (Grouping & Aggregating)** ยท ๐ŸŽจ [Visualize](https://learn-by-visuallization.org/illustrated/1-scientific-stack/pandas-groupby.html) ยท ๐Ÿ“– [PDS Handbook: Aggregation & Grouping](https://jakevdp.github.io/PythonDataScienceHandbook/03.08-aggregation-and-grouping.html) | 35m | | 12 | Cleaning: dtypes & **missing values** | [Corey Schafer Pandas](https://www.youtube.com/playlist?list=PL-osiE80TeTsWmV9i9c58mdDCSskIFdDS) โ€” **Part 9 (Cleaning Data)** ยท ๐Ÿ“– [PDS Handbook: Handling Missing Data](https://jakevdp.github.io/PythonDataScienceHandbook/03.04-missing-values.html) | 30m | | 13 | Merge / join / concat | [Kaggle Learn โ€” Pandas](https://www.kaggle.com/learn/pandas) โ€” **"Combining"** lesson ยท ๐ŸŽจ [Visualize: joins](https://learn-by-visuallization.org/illustrated/1-scientific-stack/pandas-merge.html) | 40m | | 14 | Dates & time series (for finance) | [Corey Schafer Pandas](https://www.youtube.com/playlist?list=PL-osiE80TeTsWmV9i9c58mdDCSskIFdDS) โ€” **Part 10 (Dates & Time Series)** ยท ๐Ÿ“– [pandas docs: Time series](https://pandas.pydata.org/docs/user_guide/timeseries.html) | 30m | **โœ… Checkpoint 1B** โ€” ๐Ÿ–ฅ๏ธ Local or ๐Ÿ“Š Kaggle โ€” finishes the free [Kaggle Pandas](https://www.kaggle.com/learn/pandas) cert โ€” one per topic: - **(T7)** Load a CSV; show `.head()`, `.info()`, `.describe()`; explain Series vs DataFrame. - **(T8)** Select 1 column (Series) & 2 columns (DataFrame); grab rows with `loc` (label) and `iloc` (position); set an index. - **(T9)** Filter where A > x **and** B == "y" (masks + `&` + parentheses). - **(T10)** Add a derived column, update values on a condition, drop a column. - **(T11)** `groupby` a category and `.agg` mean/count/sum of a numeric column. - **(T12)** Find missing with `.isna().sum()`; fix with `fillna`/`dropna`; correct a dtype with `astype`. - **(T13)** `merge` two DataFrames (inner vs left) and `concat` two together. - **(T14)** Parse dates with `pd.to_datetime`, set as index, resample by month. --- ## Module 1C โ€” Visualization (Matplotlib + Seaborn) | # | Topic | Open exactly this | ~Time | |---|-------|-------------------|-------| | 15 | Matplotlib anatomy: figure/axes, line plots | [Corey Schafer Matplotlib](https://www.youtube.com/playlist?list=PL-osiE80TeTvipOqomVEeZ1HRrcEvtZB_) โ€” **Part 1 (Creating & Customizing Plots)** ยท ๐Ÿ“– [PDS Handbook: Simple Line Plots](https://jakevdp.github.io/PythonDataScienceHandbook/04.01-simple-line-plots.html) | 35m | | 16 | Bar charts from CSV data | [Corey Schafer Matplotlib](https://www.youtube.com/playlist?list=PL-osiE80TeTvipOqomVEeZ1HRrcEvtZB_) โ€” **Part 2 (Bar Charts)** ยท ๐Ÿ“– [Matplotlib: Pyplot tutorial](https://matplotlib.org/stable/tutorials/pyplot.html) | 25m | | 17 | Histograms (distributions) | [Corey Schafer Matplotlib](https://www.youtube.com/playlist?list=PL-osiE80TeTvipOqomVEeZ1HRrcEvtZB_) โ€” **Part 6 (Histograms)** ยท ๐Ÿ“– [PDS Handbook: Histograms & Binnings](https://jakevdp.github.io/PythonDataScienceHandbook/04.05-histograms-and-binnings.html) | 20m | | 18 | Scatter plots (relationships) | [Corey Schafer Matplotlib](https://www.youtube.com/playlist?list=PL-osiE80TeTvipOqomVEeZ1HRrcEvtZB_) โ€” **Part 7 (Scatter Plots)** ยท ๐Ÿ“– [PDS Handbook: Simple Scatter Plots](https://jakevdp.github.io/PythonDataScienceHandbook/04.02-simple-scatter-plots.html) | 20m | | 19 | Subplots (multi-panel) | [Corey Schafer Matplotlib](https://www.youtube.com/playlist?list=PL-osiE80TeTvipOqomVEeZ1HRrcEvtZB_) โ€” **Part 10 (Subplots)** ยท ๐Ÿ“– [PDS Handbook: Multiple Subplots](https://jakevdp.github.io/PythonDataScienceHandbook/04.08-multiple-subplots.html) | 25m | | 20 | Seaborn: distributions, `heatmap`, `pairplot` | [Kaggle Learn โ€” Data Visualization](https://www.kaggle.com/learn/data-visualization) (full, free) | 2h | **โœ… Checkpoint 1C** โ€” ๐Ÿ–ฅ๏ธ Local โ€” one per topic: - **(T15)** Plot two lines on one axes with title/labels/legend; save to PNG. - **(T16)** Bar chart of a categorical count from a DataFrame. - **(T17)** Histogram of a numeric column; describe its shape (skew/spread). - **(T18)** Scatter of two columns; say whether they look correlated. - **(T19)** Put line + bar + histogram in one figure via `plt.subplots(1,3)`. - **(T20)** Seaborn `histplot`, correlation `heatmap`, and `pairplot`; read off the strongest correlated pair. --- ## Module 1D โ€” Your workbench (free compute) | # | Topic | Open exactly this | ~Time | |---|-------|-------------------|-------| | 21 | Jupyter notebooks (cells, markdown) | [Corey Schafer โ€” "Jupyter Notebook Tutorial: Introduction, Setup, and Walkthrough"](https://www.youtube.com/@coreyms/search?query=jupyter%20notebook) ยท ๐Ÿ“– [Real Python: Jupyter Notebook Intro](https://realpython.com/jupyter-notebook-introduction/) | 30m | | 22 | Google Colab (free GPU later) | [colab.research.google.com](https://colab.research.google.com) โ€” open "Welcome" notebook; *Runtime โ–ธ change to GPU* | 15m | | 23 | Kaggle Notebooks + datasets | [kaggle.com/code](https://www.kaggle.com/code) โ€” New Notebook โ–ธ Add Data | 15m | **โœ… Checkpoint 1D** โ€” one per topic: - **(T21)** ๐Ÿ–ฅ๏ธ In Jupyter: run a code cell and a markdown cell; `Shift+Enter`; restart kernel & run-all. - **(T22)** โ˜๏ธ Open the same notebook in **Colab**, switch runtime to **GPU**, confirm with `!nvidia-smi`. - **(T23)** ๐Ÿ“Š Open a **Kaggle** dataset in a Notebook and load its CSV into a DataFrame. --- ## ๐Ÿ Phase 1 capstone โ€” Real-world EDA notebook (๐Ÿ“Š Kaggle or ๐Ÿ–ฅ๏ธ Local) Pick a messy public dataset โ†’ clean (dtypes, missing) โ†’ engineer a couple of columns โ†’ **8โ€“10 labeled visualizations** โ†’ a short markdown narrative. Push the `.ipynb` to your GitHub repo. **Ready for Phase 2 when** you can take a raw CSV to a clean, well-visualized analysis end-to-end, and you know NumPy `axis` + broadcasting cold (every model uses them).