Note
Click here to download the full example code
Scorecard¶
What it's for: The Scorecard is a tabular visualization that provides a high-level overview of model performance across multiple variables, metrics, lead times, or locations. It uses color-coding to highlight areas of strength and weakness.
When to use: Use this when you have a large number of performance statistics to present simultaneously. It is particularly effective for comparing a new model version against a baseline, or for identifying which meteorological variables or forecast horizons are most problematic.
How to read: * Rows/Columns: Represent different dimensions of the evaluation (e.g., Variable vs. Forecast Lead Time). * Cell Color: Indicates the performance level. Typically, a diverging colormap is used where green represents improvement (or good performance) and red represents degradation (or poor performance) relative to a benchmark. * Interpretation: Look for patterns in the colors. For example, a whole row of red might indicate a systemic issue with a specific variable, while a column of red might indicate a drop in performance at a specific lead time.
import matplotlib.pyplot as plt
import numpy as np
import pandas as pd
from monet_plots.plots.scorecard import ScorecardPlot
from monet_plots.style import set_style
# 1. Prepare sample data in long format
np.random.seed(42) # for reproducibility
variables = ["Temperature", "Humidity", "Wind Speed", "Pressure"]
lead_times = ["+06h", "+12h", "+18h", "+24h", "+36h"]
data_list = []
for var in variables:
for lt in lead_times:
# Simulate a metric value (e.g., RMSE difference from baseline)
# Values around 0 mean similar performance, positive means worse, negative means better
metric_value = np.random.normal(loc=0, scale=0.5)
if var == "Temperature" and lt == "+06h":
metric_value = -1.2 # Example of good performance
elif var == "Pressure" and lt == "+36h":
metric_value = 1.5 # Example of poor performance
data_list.append(
{"Variable": var, "Lead Time": lt, "Metric Value": metric_value}
)
df = pd.DataFrame(data_list)
# 2. Initialize and create the plot
plot = ScorecardPlot(figsize=(10, 7))
plot.plot(
df,
x_col="Lead Time",
y_col="Variable",
val_col="Metric Value",
cmap="RdYlGn_r", # Red-Yellow-Green colormap, reversed so green is good (negative values)
center=0, # Center the colormap at 0
linewidths=0.5, # Add lines between cells
linecolor="black",
)
# 3. Add titles and labels (plot.plot sets default title and labels)
plot.ax.set_title("Model Performance Scorecard (RMSE Difference)", fontsize=14)
plot.ax.set_xlabel("Forecast Lead Time")
plot.ax.set_ylabel("Meteorological Variable")
plt.tight_layout()
# ---------------------------------------------------------
# 4. WeatherMesh-style Scorecard
# ---------------------------------------------------------
# Prepare data for WeatherMesh-style scorecard (e.g. comparing model vs observation across cities)
cities = [
"Atlanta, GA",
"Boston, MA",
"Chicago, IL",
"Houston, TX",
"Miami, FL",
"New York, NY",
]
data_wm = []
for city in cities:
for lt in [1, 3, 5, 7, 10, 14]:
mod = np.random.uniform(1, 10)
obs = np.random.uniform(1, 10)
diff = mod - obs
data_wm.append(
{"City": city, "Lead Time": lt, "Model": mod, "Obs": obs, "Diff": diff}
)
df_wm = pd.DataFrame(data_wm)
# Apply weathermesh style
set_style("weathermesh")
# Initialize and create the WeatherMesh-style plot
plot_wm = ScorecardPlot()
plot_wm.plot(
df_wm,
x_col="Lead Time",
y_col="City",
val_col="Diff",
annot_cols=["Model", "Obs"],
cbar_labels=("Model Better", "Baseline Better"),
key_text="Model RMSE | Baseline RMSE",
title="Forecast Performance in Global Cities",
cmap="RdBu_r",
center=0,
)
# Move Lead Time label to bottom center
plot_wm.ax.set_xlabel("Forecast Lead Time (Days)")
plt.show()
Total running time of the script: ( 0 minutes 0.698 seconds)
Download Python source code: plot_scorecard.py