output_view.py

from urllib.parse import unquote

import altair as alt
import pandas
import requests as req
import streamlit as st

from token_handler import init_auth_state, sendTokenRefreshMessageToParent

query_params = st.query_params

app_id = query_params.get("app_id")
batch_id = query_params.get("batch_id")
api_base_url = unquote(query_params.get("url", ""))

init_auth_state()

error = False

if app_id is None or app_id == "":
    app_id = "temp-demand-forecast"

if batch_id is None or batch_id == "":
    batch_id = "sdfsdfs"

if api_base_url == "" or api_base_url is None:
    api_base_url = "https://api.cloud.nextmv.io"

if error:
    st.stop()

headers = st.session_state.headers

results_url = f"{api_base_url}/v1/applications/{app_id}/experiments/batch/{batch_id}"
response = req.get(results_url, headers=headers)

if response.status_code != 200:
    st.error(f"Error: {response.text}")
    st.stop()
if (
    response.status_code == 403 or response.status_code == 401
) and st.session_state.get("api_key") == None:
    sendTokenRefreshMessageToParent()
    st.stop()

if response.status_code != 200:
    st.error(f"Error: {response.text}")
    st.stop()

df = pandas.DataFrame()
for summary in response.json()["grouped_distributional_summaries"]:
    summary_type = [
        "inputID",
        "instanceID",
        "versionID",
    ]  # get the distributional summaries by inputID, instanceID, and versionID
    if all(key in summary["group_keys"] for key in summary_type):
        metadata = dict(zip(summary["group_keys"], summary["group_values"]))
        values = dict(zip(summary["indicator_keys"], "indicator_distributions"))

        for indicator in summary["indicator_keys"]:
            data = {
                "inputID": metadata.get("inputID"),
                "instanceID": metadata.get("instanceID"),
                "versionID": metadata.get("versionID"),
                "indicator": indicator,
            }
            data["min"] = summary["indicator_distributions"][indicator]["min"]
            data["max"] = summary["indicator_distributions"][indicator]["max"]
            data["count"] = summary["indicator_distributions"][indicator]["count"]
            data["mean"] = summary["indicator_distributions"][indicator]["mean"]
            data["std"] = summary["indicator_distributions"][indicator]["std"]
            data["shifted_geometric_mean_value"] = summary["indicator_distributions"][
                indicator
            ]["shifted_geometric_mean"]["value"]
            data["shifted_geometric_mean_shift"] = summary["indicator_distributions"][
                indicator
            ]["shifted_geometric_mean"]["shift"]
            data["p01"] = summary["indicator_distributions"][indicator]["percentiles"][
                "p01"
            ]
            data["p05"] = summary["indicator_distributions"][indicator]["percentiles"][
                "p05"
            ]
            data["p10"] = summary["indicator_distributions"][indicator]["percentiles"][
                "p10"
            ]
            data["p25"] = summary["indicator_distributions"][indicator]["percentiles"][
                "p25"
            ]
            data["p50"] = summary["indicator_distributions"][indicator]["percentiles"][
                "p50"
            ]
            data["p75"] = summary["indicator_distributions"][indicator]["percentiles"][
                "p75"
            ]
            data["p90"] = summary["indicator_distributions"][indicator]["percentiles"][
                "p90"
            ]
            data["p95"] = summary["indicator_distributions"][indicator]["percentiles"][
                "p95"
            ]
            data["p99"] = summary["indicator_distributions"][indicator]["percentiles"][
                "p99"
            ]
            df = pandas.concat([df, pandas.DataFrame([data])], ignore_index=True)

indicators = df["indicator"].unique()
columns = [
    col
    for col in df.columns
    if col not in ["inputID", "instanceID", "versionID", "indicator"]
]

# Create a dropdown menu for the user to select the indicator and column
st.selectbox("Select a metric:", indicators, key="selected_indicator")
st.selectbox("Select a statistic:", columns, key="selected_column")

# # Initialize session_state if it doesn't exist
# if "selected_indicator" not in st.session_state:
#     selected_indicator = indicators[0]

# if "selected_column" not in st.session_state:
#     selected_column = columns[0]

df_filtered = df[df["indicator"] == st.session_state.selected_indicator]
chart = (
    alt.Chart(df_filtered)
    .mark_bar()
    .encode(
        x="inputID:N",
        y=alt.Y(st.session_state.selected_column, stack="zero"),
        color="instanceID:N",
        tooltip=[st.session_state.selected_column],
    )
    .properties(width=800, height=400)
)

st.altair_chart(chart)

# if selected_column == "mean" and all(df_filtered["std"] > 1):
#     st.write(df_filtered)
#     df_filtered["lower_bound"] = df_filtered["mean"] - 1.96 * df_filtered["std"]
#     df_filtered["upper_bound"] = df_filtered["mean"] + 1.96 * df_filtered["std"]
#     st.write(df_filtered)

#     base = alt.Chart(df_filtered).encode(
#         x="instanceID:N",
#         color="instanceID:N",
#     )

#     bars = base.mark_bar().encode(
#         y=alt.Y(selected_column, title=selected_column),
#     )

#     error_bars = base.mark_errorbar().encode(
#         y="lower_bound:Q",
#         y2="upper_bound:Q",
#     )

#     chart = (
#         (bars + error_bars)
#         .properties(width=800)
#         .facet(
#             column="inputID",
#         )
#         .configure_axis(
#             labelFontSize=15,
#             titleFontSize=15,
#         )
#         .configure_title(fontSize=25)
#     )
# else:
# base = alt.Chart(df_filtered).encode(
#     x="instanceID:N",
#     color="instanceID:N",
# )

# bars = base.mark_bar().encode(
#     y=alt.Y(selected_column, title=selected_column),
# )
# chart = (
#     bars.properties(width=800)
#     .facet(
#         column="inputID",
#     )
#     .configure_axis(
#         labelFontSize=15,
#         titleFontSize=15,
#     )
#     .configure_title(fontSize=25)
# )

# st.altair_chart(chart)

chart = (
    alt.Chart(df_filtered)
    .mark_rect()
    .encode(
        x="inputID:N",
        y="instanceID:N",
        color=alt.Color(
            st.session_state.selected_column,
            scale=alt.Scale(scheme="blues", type="log"),
        ),
        tooltip=[st.session_state.selected_column],
    )
    .properties(width=800, height=400)
)

st.altair_chart(chart)