Your first dataset - Docs

If you want to see what imibare gives you, load a dataset and inspect it. This tutorial installs the package, reads Rwanda’s monthly Consumer Price Index (CPI), and plots the general index.

Before you start

You need Python 3.11 or later and pip, which comes with most Python installations. We install matplotlib with the package because you will use it for the final chart.

Install the package

Run this in a terminal:

pip install imibare matplotlib

The install is complete when pip reports that it installed imibare. You only need to run this command once for an environment.

Load the Consumer Price Index

Create a Python file, or use an interactive Python session, and run:

import imibare as imi

df = imi.load("rw.nisr.cpi.monthly")
print(df.head())

imi.load() returns a pandas DataFrame. It downloads the current Parquet file from the API on the first call, then keeps a local copy under ~/.imibare/cache/ so another call within an hour does not download it again.

The first rows will look broadly like this:

        date                         category       value
0 2009-02-01  alcoholic_beverages_and_tobacco   87.380474
1 2009-02-01                bread_and_cereals   92.058744
2 2009-02-01            clothing_and_footwear   83.509793

The exact newest rows change when NISR publishes an update. The column names and the meaning of the data stay documented on the dataset page.

Check what you loaded

Before plotting a dataset, inspect its shape and the values that identify each series. This CPI dataset is in long format: 1 row for each category in each month.

print(df.dtypes)
print(df.shape)
print(df["category"].nunique())
print(sorted(df["category"].unique())[:4])

You should find 3 columns: date, category, and value. There are 18 CPI categories, so the row count grows by 18 whenever NISR adds a month; pandas 2 may display category as object rather than str.

Plot the general index

The headline CPI series is general_index_(cpi). Filter to it before plotting, because the other 17 categories are components rather than duplicate observations of the same series.

import matplotlib.pyplot as plt

general = df[df["category"] == "general_index_(cpi)"].sort_values("date")

plt.figure(figsize=(12, 4))
plt.plot(general["date"], general["value"], linewidth=1.5)
plt.title("Rwanda CPI, general index (2014 = 100)")
plt.xlabel("Date")
plt.ylabel("Index value")
plt.tight_layout()
plt.savefig("rw_cpi.png", dpi=150)
plt.show()

This saves rw_cpi.png and opens the chart when your environment supports it. The line stays near 100 through 2014 because the series uses 2014 as its base year, so a value of 241 means prices were about 2.4 times their 2014 level.

Find another dataset

Use imi.catalog() when you know the country but not the ID. The returned records include the ID, institution, coverage, columns, source, and pipeline label for each dataset.

datasets = imi.catalog(country="RW")

for dataset in datasets[:5]:
    print(f"{dataset.id:45s}  {dataset.name}")

Change country="RW" to country="KE" for Kenya. You can also browse the full catalog and copy an ID from a dataset page.

Keep going

You now have a real DataFrame, a chart, and a way to find the next dataset. These are the common next steps:

Load exchange rates: imi.load("rw.bnr.fx.daily")
Load Kenyan CPI: imi.load("ke.knbs.cpi.monthly")
Limit a date range: imi.load("rw.nisr.cpi.monthly", start="2020-01", end="2023-12")
Return a Polars DataFrame: imi.load("rw.nisr.cpi.monthly", engine="polars")
Check every parameter: Python package reference

If you prefer SQL, DuckDB can read the same file directly. The DuckDB guide shows the URL pattern and several queries.