Datasets
Explore and use Datadex datasets in your preferred tools!
π Explore
You can a sense of how the datasets produced look by exploring them in HuggingFace. With each commit to main
, Datadex will push a new version of the datasets as Parquet files.
These are the available Datasets!
π§ Use
Since the datasets are just Parquet files somewhere. You can use pretty much any tool or framework to explore them. Letβs look at the Spain IPC dataset with Polars.
import polars as pl
= pl.read_parquet(
df "https://huggingface.co/datasets/datonic/spain_ipc/resolve/refs%2Fconvert%2Fparquet/default/main/0000.parquet"
)
4) df.sample(
shape: (4, 6)
periodo | clases | indice | variacion_mensual | variacion_anual | variacion_en_lo_que_va_de_ano |
---|---|---|---|---|---|
date | str | f64 | f64 | f64 | f64 |
2018-05-01 | "1254 Seguros relacionados con β¦ | 96.323 | 0.0 | 1.5 | 0.0 |
2010-04-01 | "0312 Prendas de vestir" | 98.85 | 10.3 | -1.2 | -4.8 |
2014-08-01 | "0952 Prensa" | 86.534 | 0.6 | 1.7 | 0.9 |
2013-08-01 | "0942 Servicios culturales" | 99.054 | 0.3 | 10.7 | 1.4 |
= df.filter(pl.col("clases") == "0952 Prensa") ipc_prensa
import altair as alt
alt.Chart(ipc_prensa).mark_line().encode(=alt.X("periodo", title="Period", axis=alt.Axis(labelAngle=-45)),
x=alt.Y("indice", title="Index"),
y=["periodo", "indice"],
tooltip="container", height=400, title="IPC Prensa Over Time").interactive() ).properties(width