forecast_precip_probability
rain in the forecast
12.50at prediction · 56th percentile
- min
- 0.00
- p25
- 0.00
- median
- 8.33
- p75
- 41.67
- max
- 100.00
- mean
- 23.07
- std
- 28.24
- p05
- 0.00
- p95
- 79.17
- n
- 1077
train tomorrow
Training status for
Prediction generated Oct 9, 2026, 12:07 AM PDT
Train probability
59%
Likely training day
Predicted effort 27.9relative-effort points
Model says likely session (59%). Credit regular train rate for this day of the week [value=0.321 fraction 0-1 (typical 0.173 fraction 0-1, 96% of history is at or below)]. Watch out for the temp high forecast [value=56.84 °F (typical 58.46 °F, 47% of history is at or below)]. Target: ~28 relative-effort points.
How the estimate moves
Starting from the model's average day (40.2%), each row applies the next SHAP-IQ effect to reach 59.1%.
Feature rows show main effects after separating pairwise interactions; interaction rows show the pair effect.
Local scale: 39% to 60%
Model averageFinal prediction
regular train rate for this day of the week+6.6 points · 46.8%helping · +0.270 log odds
SHAP-IQ contribution +0.270 log odds; probability change +6.6 points.
SHAP-IQ log odds vs. feature value. Above 0 pushes toward training; below 0 pushes away. Today: value 0.3, +0.270 log odds.
Historical days: 19. Feature values ranged from -0.0 to 0.4. SHAP-IQ log-odds ranged from -0.761 log odds to +0.365 log odds.
-0.0feature value0.4SHAP-IQ log-odds range -0.761 log odds to +0.365 log odds
how often you've trained this past week+4.7 points · 51.5%helping · +0.187 log odds
SHAP-IQ contribution +0.187 log odds; probability change +4.7 points.
SHAP-IQ log odds vs. feature value. Above 0 pushes toward training; below 0 pushes away. Today: value 5.0, +0.188 log odds.
Historical days: 5. Feature values ranged from -0.0 to 5.8. SHAP-IQ log-odds ranged from -0.109 log odds to +0.209 log odds.
-0.0feature value5.8SHAP-IQ log-odds range -0.109 log odds to +0.209 log odds
regular train rate for this month+4.6 points · 56.2%helping · +0.186 log odds
SHAP-IQ contribution +0.186 log odds; probability change +4.6 points.
SHAP-IQ log odds vs. feature value. Above 0 pushes toward training; below 0 pushes away. Today: value 0.3, +0.186 log odds.
Historical days: 18. Feature values ranged from -0.1 to 0.8. SHAP-IQ log-odds ranged from -0.371 log odds to +0.242 log odds.
-0.1feature value0.8SHAP-IQ log-odds range -0.371 log odds to +0.242 log odds
the temp high forecast-4.2 points · 51.9%hurting · -0.170 log odds
SHAP-IQ contribution -0.170 log odds; probability change -4.2 points.
SHAP-IQ log odds vs. feature value. Above 0 pushes toward training; below 0 pushes away. Today: value 13.8, -0.170 log odds.
Historical days: 24. Feature values ranged from 1.4 to 32.5. SHAP-IQ log-odds ranged from -0.347 log odds to +0.298 log odds.
1.4feature value32.5SHAP-IQ log-odds range -0.347 log odds to +0.298 log odds
your training frequency over the past month-3.0 points · 48.9%hurting · -0.121 log odds
SHAP-IQ contribution -0.121 log odds; probability change -3.0 points.
SHAP-IQ log odds vs. feature value. Above 0 pushes toward training; below 0 pushes away. Today: value 18.0, -0.120 log odds.
Historical days: 14. Feature values ranged from -0.3 to 20.2. SHAP-IQ log-odds ranged from -0.141 log odds to +0.139 log odds.
-0.3feature value20.2SHAP-IQ log-odds range -0.141 log odds to +0.139 log odds
longer-term fitness base+3.0 points · 51.9%helping · +0.120 log odds
SHAP-IQ contribution +0.120 log odds; probability change +3.0 points.
SHAP-IQ log odds vs. feature value. Above 0 pushes toward training; below 0 pushes away. Today: value 46.0, +0.120 log odds.
Historical days: 19. Feature values ranged from -4.1 to 63.4. SHAP-IQ log-odds ranged from -0.173 log odds to +0.206 log odds.
-4.1feature value63.4SHAP-IQ log-odds range -0.173 log odds to +0.206 log odds
time since your last hard effort+2.0 points · 54.0%helping · +0.082 log odds
SHAP-IQ contribution +0.082 log odds; probability change +2.0 points.
SHAP-IQ log odds vs. feature value. Above 0 pushes toward training; below 0 pushes away. Today: value 0.0, +0.082 log odds.
Historical days: 14. Feature values ranged from -2.4 to 31.8. SHAP-IQ log-odds ranged from -0.219 log odds to +0.205 log odds.
-2.4feature value31.8SHAP-IQ log-odds range -0.219 log odds to +0.205 log odds
how the temp high compares to normal for this time of year+1.6 points · 55.5%helping · +0.064 log odds
SHAP-IQ contribution +0.064 log odds; probability change +1.6 points.
SHAP-IQ log odds vs. feature value. Above 0 pushes toward training; below 0 pushes away. Today: value -3.3, +0.064 log odds.
Historical days: 24. Feature values ranged from -7.7 to 9.8. SHAP-IQ log-odds ranged from -0.202 log odds to +0.237 log odds.
-7.7feature value9.8SHAP-IQ log-odds range -0.202 log odds to +0.237 log odds
recent training load in your legs+1.3 points · 56.8%helping · +0.052 log odds
SHAP-IQ contribution +0.052 log odds; probability change +1.3 points.
SHAP-IQ log odds vs. feature value. Above 0 pushes toward training; below 0 pushes away. Today: value 67.1, +0.052 log odds.
Historical days: 19. Feature values ranged from -6.8 to 93.2. SHAP-IQ log-odds ranged from -0.065 log odds to +0.075 log odds.
-6.8feature value93.2SHAP-IQ log-odds range -0.065 log odds to +0.075 log odds
interaction between regular train rate for this month and the temp high forecast-1.2 points · 55.6%hurting · -0.050 log odds
SHAP-IQ contribution -0.050 log odds; probability change -1.2 points.
all other features+3.5 points · 59.1%helping · +0.144 log odds
SHAP-IQ contribution +0.144 log odds; probability change +3.5 points.
Every model feature, ordered by model feature importance. The amber line is the value used for this prediction; the violin shows the training-data distribution with dashed p25/p75 and solid median.
importance 0.095
trained_days_7
how often you've trained this past week
5.00at prediction · 97th percentile
- min
- 0.00
- p25
- 0.00
- median
- 2.00
- p75
- 3.00
- max
- 7.00
- mean
- 2.04
- std
- 1.74
- p05
- 0.00
- p95
- 5.00
- n
- 1094
forecast_temp_high
the temp high forecast
13.80at prediction · 47th percentile
- min
- -3.90
- p25
- 9.80
- median
- 14.70
- p75
- 21.00
- max
- 34.50
- mean
- 15.58
- std
- 7.09
- p05
- 5.60
- p95
- 27.84
- n
- 1077
dow_train_rate
regular train rate for this day of the week
0.32at prediction · 96th percentile
- min
- 0.00
- p25
- 0.03
- median
- 0.17
- p75
- 0.26
- max
- 0.50
- mean
- 0.17
- std
- 0.12
- p05
- 0.00
- p95
- 0.32
- n
- 1094
forecast_precip_evening
rain chances during your evening window
0.00at prediction · 63th percentile
- min
- 0.00
- p25
- 0.00
- median
- 0.00
- p75
- 40.00
- max
- 100.00
- mean
- 23.29
- std
- 35.44
- p05
- 0.00
- p95
- 100.00
- n
- 1077
month_train_rate
regular train rate for this month
0.27at prediction · 66th percentile
- min
- 0.00
- p25
- 0.00
- median
- 0.14
- p75
- 0.42
- max
- 1.00
- mean
- 0.22
- std
- 0.22
- p05
- 0.00
- p95
- 0.60
- n
- 1094
chronic_load_28
longer-term fitness base
46.03at prediction · 93th percentile
- min
- 0.00
- p25
- 9.01
- median
- 21.64
- p75
- 31.46
- max
- 78.34
- mean
- 21.27
- std
- 16.17
- p05
- 0.00
- p95
- 49.05
- n
- 1094
days_since_last_hard
time since your last hard effort
0.00at prediction · 15th percentile
- min
- 0.00
- p25
- 1.00
- median
- 4.00
- p75
- 8.00
- max
- 39.00
- mean
- 6.12
- std
- 7.04
- p05
- 0.00
- p95
- 21.00
- n
- 841
forecast_temp_high_vs_seasonal
how the temp high compares to normal for this time of year
-3.27at prediction · 12th percentile
- min
- -11.15
- p25
- -1.84
- median
- 0.15
- p75
- 2.58
- max
- 14.48
- mean
- 0.42
- std
- 3.42
- p05
- -4.73
- p95
- 6.30
- n
- 1077
forecast_temp_low_vs_seasonal
how the temp low compares to normal for this time of year
0.68at prediction · 55th percentile
- min
- -13.35
- p25
- -1.41
- median
- 0.35
- p75
- 2.06
- max
- 9.35
- mean
- 0.32
- std
- 2.77
- p05
- -4.32
- p95
- 4.79
- n
- 1077
trained_days_30
your training frequency over the past month
18.00at prediction · 98th percentile
- min
- 0.00
- p25
- 4.00
- median
- 8.00
- p75
- 14.00
- max
- 21.00
- mean
- 8.54
- std
- 5.97
- p05
- 0.00
- p95
- 17.00
- n
- 1094
acute_load_7
recent training load in your legs
67.09at prediction · 95th percentile
- min
- 0.00
- p25
- 1.97
- median
- 16.39
- p75
- 32.28
- max
- 144.73
- mean
- 21.63
- std
- 22.33
- p05
- 0.00
- p95
- 65.67
- n
- 1094
days_since_last_long_ride
time since your last really long session
0.00at prediction · 12th percentile
- min
- 0.00
- p25
- 2.00
- median
- 5.00
- p75
- 11.00
- max
- 42.00
- mean
- 7.92
- std
- 7.97
- p05
- 0.00
- p95
- 25.00
- n
- 821
mileage_chronic_28
your longer-term mileage base
55408.61at prediction · 100th percentile
- min
- 114.62
- p25
- 6422.64
- median
- 14636.66
- p75
- 26718.47
- max
- 51880.25
- mean
- 16697.81
- std
- 12618.61
- p05
- 384.30
- p95
- 39279.40
- n
- 1094
streak_length
your current streak momentum
2.00at prediction · 94th percentile
- min
- 0.00
- p25
- 0.00
- median
- 0.00
- p75
- 1.00
- max
- 9.00
- mean
- 0.55
- std
- 1.17
- p05
- 0.00
- p95
- 3.00
- n
- 1094
trained_days_2
how much you've trained the past 2 days
2.00at prediction · 100th percentile
- min
- 0.00
- p25
- 0.00
- median
- 0.00
- p75
- 1.00
- max
- 2.00
- mean
- 0.59
- std
- 0.70
- p05
- 0.00
- p95
- 2.00
- n
- 1094
These metrics come from a quick holdout of the most recent 365 days (2025-10-08 to 2026-10-07), scored by a model trained only on the 729 earlier days. The deployed model is then retrained on all data, including the holdout, because recency matters; its true accuracy is not measured here.
Will I train? (classifier)
AUC
0.655
0.5 = coin flip
Accuracy
67.4%
always-majority baseline 62.2%
Precision
67.9%
Recall
26.1%
F1
0.377
Log loss
0.626
lower is better
Brier
0.218
lower is better
Train rate in holdout
37.8%
False-positive rate (x) vs. true-positive rate (y); above the dashed line beats chance.
Mean predicted probability (x) vs. observed train rate (y); dot area scales with n.
How hard? (effort regressor)
MAE
36.9
relative-effort points
Median baseline MAE
35.3
always predict the training median
Holdout rows
138
184 train rows (training days only)
Coming soon
Prediction impact tracking will be released once there is more data.