In [1]:
import plotly.express as px
import pandas as pd
In [2]:
# df = pd.read_parquet("era5-pds/measurements-m1.parquet")
df = pd.read_parquet("era5-pds/measurements-i10k.parquet")
# df = pd.read_parquet("era5-pds/measurements-ryzen3.parquet")
# df = pd.read_parquet("era5-pds/measurements-i13k.parquet")
df = df.query("clevel > 0")  # get rid of no compression results
In [3]:
category_orders = {"dset": ["flux", "wind", "pressure", "precip", "snow"],
                   "filter": ["nofilter", "shuffle", "bitshuffle", "bytedelta"]}
labels = {
    "cratio": "Compression ratio (x times)",
    "cspeed": "Compression speed (GB/s)",
    "dspeed": "Decompression speed (GB/s)",
    "codec": "Codec",
    "dset": "Dataset",
    "filter": "Filter",
    "cratio * cspeed": "Compression ratio x Compression speed",
    "cratio * dspeed": "Compression ratio x Decompression speed",
    "cratio * cspeed * dspeed": "Compression ratio x Compression x Decompression speeds",
    }
In [4]:
hover_data = {"filter": False, "codec": True, "cratio": ':.1f', "cspeed": ':.2f',
              "dspeed": ':.2f', "dset": True, "clevel": True}
fig = px.box(df, x="cratio", color="filter", points="all", hover_data=hover_data,
             labels=labels, range_x=(0, 60), range_y=(-.4, .35),)
fig.update_layout(
    title={
        'text': "Compression ratio vs filter (larger is better)",
        #'y':0.9,
        'x':0.25,
        'xanchor': 'left',
        #'yanchor': 'top'
    },
    #xaxis_title="Filter",
)
fig.show()