A terminal viewer for Jupyter notebooks. It’s like cat for ipynb files.
% nbpreview --theme material --image-drawing braille notebook.ipynb
nbpreview
─────────────────────────────────────────────────────────────────────────
╭────────────────────────────────────────────────────────────────────────╮
[1]: │ from typing import Optional │
│ │
│ import arviz as az │
│ import matplotlib.pyplot as plt │
│ import pandas as pd │
│ import pymc as pm │
│ from arviz import InferenceData │
│ from matplotlib.axes import Axes │
│ from matplotlib.axes._subplots import Subplot │
│ from pandas import DataFrame │
│ │
│ import plots │
╰────────────────────────────────────────────────────────────────────────╯
Thanks for checking out nbpreview. This example notebook is inspired by
an example in Bayesian Analysis with Python by Osvaldo Martin. A more
detailed breakdown of how nbpreview renders notebooks, examples, and
command-line options may be found in the documentation.
## Load data
─────────────────────────────────────────────────────────────────────────
This dataset contains the heights (Length) and age (Month) of newborn
girls.
╭────────────────────────────────────────────────────────────────────────╮
[2]: │ babies_data = pd.read_csv( │
│ "https://github.com/aloctavodia/BAP/blob/master/code/data/babies.… │
│ ).rename(columns={"Lenght": "Length"}) │
│ months_of_interest = list(range(0, 13, 4)) │
│ ( │
│ babies_data.groupby("Month") │
│ .agg( │
│ mean_length=("Length", "mean"), │
│ median_length=("Length", "median"), │
│ mean_std=("Length", "std"), │
│ measurement_count=("Month", "count"), │
│ ) │
│ .loc[months_of_interest] │
│ ) │
╰────────────────────────────────────────────────────────────────────────╯
[2]: 🌐 Click to view HTML
[2]: mean_length median_length mean_std measurement_count
Month
────────────────────────────────────────────────────────────────────
0 49.458333 49.25 1.824285 48
4 62.060606 62.50 2.548767 33
8 68.809524 69.00 2.677714 42
12 74.456522 74.50 2.549122 23
╭────────────────────────────────────────────────────────────────────────╮
[3]: │ babies_data.plot.scatter( │
│ x="Month", │
│ y="Length", │
│ figsize=(30, 7), │
│ s=500, │
│ xticks=[], │
│ yticks=[], │
│ ); │
╰────────────────────────────────────────────────────────────────────────╯
🖼 Click to view Image
⣿⡏⠉⠉⠉⠉⠉⠉⠉⠉⠉⠉⠉⠉⠉⠉⠉⠉⠉⠉⠉⠉⠉⠉⠉⠉⠉⠉⠉⠉⠉⠉⠉⠉⠉⠉⠉⠉⠉⠉⠉⠉⠉⠉⠉⠉⠉⠉⠉⠉⠉⠉⠉⠉⠉⢉⡉⠉⣉⠉⠉⣯⠉⢩⡍⠉⢉⡉⠉⣿⠉⠉⢹
⣿⡇⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⣀⠀⠀⡀⠀⢠⡄⠀⢀⠀⠀⣦⠀⢸⡇⠀⣾⠀⠀⣿⠀⢸⡇⠀⢸⡇⠀⣿⠀⠀⢸
⣿⡇⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⡀⠀⠀⠀⠀⣀⠀⢀⣄⠀⢸⡇⠀⣿⠀⠀⣿⠀⢸⡇⠀⣿⠀⠀⡿⠀⢸⡇⠀⣻⠂⠀⣿⠀⢸⡇⠀⠸⠇⠀⠿⠀⠀⢸
⣿⡇⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⢀⡀⠀⣠⡄⠀⣶⠀⢸⡇⠀⢸⡇⠀⣿⠀⠸⡇⠀⢸⡇⠀⣿⠀⠀⡿⠀⢸⡇⠀⠈⠀⠀⠉⠀⠘⠃⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⢸
⣿⡇⠀⠀⠀⠀⠀⠀⠀⠀⡀⠀⢀⡀⠀⣲⠀⠀⣷⠀⢸⡇⠀⣿⡇⠀⣿⠀⠸⡇⠀⠘⠃⠀⠛⠀⠐⠆⠀⠈⠁⠀⠈⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⢸
⣿⡇⠀⠀⠀⠀⠀⢶⠀⠀⣿⠀⢸⡇⠀⣿⠀⠀⠛⠀⠘⠃⠀⠸⠅⠀⠉⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⢸
⣿⡇⠀⠀⢰⡆⠀⣿⠀⠀⠿⠀⠈⠁⠀⠙⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⢸
⣿⡇⠀⠀⢸⡇⠀⠈⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⢸
⣿⣷⣶⣶⣶⣶⣶⣶⣶⣶⣶⣶⣶⣶⣶⣶⣶⣶⣶⣶⣶⣶⣶⣶⣶⣶⣶⣶⣶⣶⣶⣶⣶⣶⣶⣶⣶⣶⣶⣶⣶⣶⣶⣶⣶⣶⣶⣶⣶⣶⣶⣶⣶⣶⣶⣶⣶⣶⣶⣶⣶⣶⣶⣶⣶⣶⣶⣶⣶⣶⣶⣶⣾
## Make model
─────────────────────────────────────────────────────────────────────────
We'll model Length as a linear function of the square root of Month.
Length ∼𝒩(μ= α+ β√(Month), ϵ)
Where the variance itself is also a linear function of Month.
ϵ∼γ+ σMonth
╭────────────────────────────────────────────────────────────────────────╮
[4]: │ with pm.Model(coords={"time_idx": babies_data.index}) as babies_model: │
│ # Priors │
│ alpha = pm.Normal("alpha", sigma=10) │
│ beta = pm.Normal("beta", sigma=10) │
│ gamma = pm.HalfNormal("gamma", sigma=10) │
│ sigma = pm.HalfNormal("sigma", sigma=10) │
│ │
│ month = pm.MutableData( │
│ "month", │
│ value=babies_data["Month"].astype(float), │
│ ) │
│ │
│ mu = pm.Deterministic( │
│ "mu", │
│ alpha + beta * month ** 0.5, │
│ dims="time_idx", │
│ ) │
│ epsilon = pm.Deterministic( │
│ "epsilon", │
│ gamma + sigma * month, │
│ dims="time_idx", │
│ ) │
│ pm.Normal( │
│ "length", │
│ mu=mu, │
│ sigma=epsilon, │
│ observed=babies_data["Length"], │
│ dims="time_idx", │
│ ) │
│ │
│ # Sample model │
│ babies_idata = pm.sample(tune=2_000, return_inferencedata=True) │
│ babies_idata.extend(pm.sample_posterior_predictive(babies_idata)) │
╰────────────────────────────────────────────────────────────────────────╯
Auto-assigning NUTS sampler...
Initializing NUTS using jitter+adapt_diag...
Multiprocess sampling (4 chains in 4 jobs)
NUTS: [alpha, beta, gamma, sigma]
🌐 Click to view HTML
100.00% [12000/12000 00:07<00:00 Sampling 4 chains, 0 divergences]
Sampling 4 chains for 2_000 tune and 1_000 draw iterations (8_000 +
4_000 draws total) took 20 seconds.
🌐 Click to view HTML
100.00% [4000/4000 00:00<00:00]
╭────────────────────────────────────────────────────────────────────────╮
[5]: │ epsilon │
╰────────────────────────────────────────────────────────────────────────╯
[5]: epsilon∼Deterministic(f(gamma, sigma))
## Plots
─────────────────────────────────────────────────────────────────────────
Let's plot the fit as a band where we expect 95% of the data to be
contained in.
╭────────────────────────────────────────────────────────────────────────╮
[6]: │ _, hdi_ax = plt.subplots(figsize=(25, 5)) │
│ az.plot_hdi( │
│ x=babies_data["Month"], │
│ y=babies_idata["posterior_predictive"]["length"], │
│ hdi_prob=0.95, │
│ fill_kwargs={"alpha": 1.0}, │
│ ax=hdi_ax, │
│ ).set(xticks=[], yticks=[]); │
╰────────────────────────────────────────────────────────────────────────╯
🖼 Click to view Image
⡏⠉⠉⠉⠉⠉⠉⠉⠉⠉⠉⠉⠉⠉⠉⠉⠉⠉⠉⠉⠉⠉⠉⠉⠉⠉⠉⠉⠉⠉⠉⠉⠉⠉⠉⠉⠉⠉⠉⠉⠉⠉⠉⠉⠉⠉⠉⠉⠉⠉⠉⠉⠉⠉⠉⠉⠉⠉⠉⢉⣉⣉⣉⣉⣉⣩⣭⣭⣭⡍⠉⠉⢹
⡇⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⢀⣀⣀⣀⣀⣤⣤⣤⣤⣴⣶⣶⣶⣶⣶⣿⣿⣿⣿⣿⣿⣿⣿⣿⣿⣿⣿⣿⣿⡇⠀⠀⢸
⡇⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⣀⣀⣀⣠⣤⣤⣤⣶⣶⣶⣶⣿⣿⣿⣿⣿⣿⣿⣿⣿⣿⣿⣿⣿⣿⣿⣿⣿⠿⠿⠿⠿⠿⠟⠛⠛⠛⠛⠛⠛⠛⠉⠉⠉⠁⠀⠀⢸
⡇⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⢀⣀⣀⣤⣤⣴⣶⣶⣾⣿⣿⣿⣿⣿⣿⣿⣿⣿⣿⡿⠿⠿⠿⠿⠛⠛⠛⠛⠛⠉⠉⠉⠉⠉⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⢸
⡇⠀⠀⠀⠀⠀⠀⣀⣠⣤⣶⣾⣿⣿⣿⣿⣿⣿⠿⠿⠿⠛⠛⠛⠋⠉⠉⠉⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⢸
⡇⠀⠀⢀⣤⣶⣿⣿⣿⠿⠟⠛⠋⠉⠉⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⢸
⡇⠀⠀⢸⠿⠛⠋⠁⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⢸
Let's directly compare the distributions of lengths for newborns that are
0, 4, 8, and 12 months old.
╭────────────────────────────────────────────────────────────────────────╮
[7]: │ _, dist_ax = plt.subplots(figsize=(25, 5)) │
│ │
│ │
│ def plot_length_dist( │
│ babies_idata: InferenceData, │
│ babies_data: DataFrame, │
│ month: int, │
│ ax: Optional[Axes] = None, │
│ color: Optional[str] = None, │
│ ) -> Subplot: │
│ """Plot the length distribution given an age in months.""" │
│ if ax is None: │
│ ax = plt.gca() │
│ │
│ length_data = babies_idata.sel( │
│ time_idx=babies_data.loc[lambda df: df["Month"] == month].ind… │
│ )["posterior_predictive"].stack( │
│ dim=[ │
│ "chain", │
│ "draw", │
│ "time_idx", │
│ ] │
│ )[ │
│ "length" │
│ ] │
│ plot = az.plot_dist( │
│ length_data, │
│ fill_kwargs={"alpha": 1}, │
│ ax=ax, │
│ color=color, │
│ ) │
│ return plot │
│ │
│ │
│ for idx, month in enumerate(months_of_interest): │
│ color = f"C{idx}" │
│ plot_length_dist( │
│ babies_idata, │
│ babies_data=babies_data, │
│ month=month, │
│ color=color, │
│ ax=dist_ax, │
│ ).set(xticks=[], yticks=[]); │
╰────────────────────────────────────────────────────────────────────────╯
🖼 Click to view Image
⡏⠉⠉⠉⠉⠉⠉⠉⠉⠉⠉⠉⠉⠉⠉⠉⣩⣭⡉⠉⠉⠉⠉⠉⠉⠉⠉⠉⠉⠉⠉⠉⠉⠉⠉⠉⠉⢉⣉⠉⠉⠉⠉⠉⠉⠉⠉⠉⠉⠉⠉⠉⠉⠉⠉⠉⠉⠉⠉⠉⠉⠉⠉⠉⠉⠉⠉⠉⠉⠉⠉⠉⢹
⡇⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⣰⣿⣿⣿⡆⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⣰⣿⣿⣷⡀⠀⠀⠀⠀⣰⣾⣶⡄⠀⠀⠀⣠⣤⣀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⢸
⡇⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⢰⣿⣿⣿⣿⣿⡄⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⣰⣿⣿⣿⣿⣷⡀⠀⠀⣼⣿⣿⣿⣿⣄⢀⣾⣿⣿⣿⣷⡀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⢸
⡇⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⢠⣿⣿⣿⣿⣿⣿⣷⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⣰⣿⣿⣿⣿⣿⣿⣧⠀⣼⣿⣿⣿⣿⣿⣿⣾⣿⣿⣿⣿⣿⣷⡀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⢸
⡇⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⢀⣾⣿⣿⣿⣿⣿⣿⣿⣧⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⢠⣿⣿⣿⣿⣿⣿⣿⣿⣿⣿⣿⣿⣿⣿⣿⣿⣿⣿⣿⣿⣿⣿⣿⣿⣄⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⢸
⡇⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⢠⣾⣿⣿⣿⣿⣿⣿⣿⣿⣿⣧⡀⠀⠀⠀⠀⠀⠀⠀⠀⣰⣿⣿⣿⣿⣿⣿⣿⣿⣿⣿⣿⣿⣿⣿⣿⣿⣿⣿⣿⣿⣿⣿⣿⣿⣿⣿⣦⡀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⢸
⡇⠀⠀⠀⠀⠀⠀⢀⣀⣠⣴⣿⣿⣿⣿⣿⣿⣿⣿⣿⣿⣿⣿⣿⣦⣀⣀⠀⣀⣀⣤⣾⣿⣿⣿⣿⣿⣿⣿⣿⣿⣿⣿⣿⣿⣿⣿⣿⣿⣿⣿⣿⣿⣿⣿⣿⣿⣿⣿⣿⣿⣦⣄⣀⣀⠀⠀⠀⠀⠀⠀⠀⢸
