Add type filtering, drop hard-coded filters and route filter, tune backtest

2 files changed, 32 insertions, 18 deletions

diff --git a/cli/cli.go b/cli/cli.go
index bd54b35..e2138cd 100644
--- a/cli/cli.go
+++ b/cli/cli.go
@@ -28,11 +28,6 @@ var (
 	enablePprof = flag.Bool("pprof", false, "expose pprof endpoints at /debug/pprof")
 )
 
-const (
-	backtestMinimumDistanceM   = 10_000
-	backtestMinimumMovingRatio = 0.8
-)
-
 func Run(db *sql.DB, configPath string, appConfig config.Config) {
 	switch {
 	case *fetchOSM:
@@ -127,16 +122,21 @@ func runBacktest(db *sql.DB) error {
 		fmt.Fprintf(table, "p%g\t%.1f\n", p, distribution[p])
 	}
 	table.Flush()
-	slog.Info("Moving time backtest", "rides", len(samples), "min_distance_m", backtestMinimumDistanceM, "min_moving_ratio", backtestMinimumMovingRatio, "mean_err_pct", math.Round(meanAbsErr(absErrors)*10)/10, "stderr_pct", math.Round(stdErrAbsErr(absErrors)*10)/10)
+	slog.Info("Moving time backtest", "rides", len(samples), "mean_err_pct", math.Round(meanAbsErr(absErrors)*10)/10, "stderr_pct", math.Round(stdErrAbsErr(absErrors)*10)/10)
 	return nil
 }
 
 func isBacktestEligible(item ride.StoredRide) bool {
-	return item.GPXPath != "" &&
-		item.DistanceM >= backtestMinimumDistanceM &&
-		item.ExternalMovingTimeS > 0 &&
-		item.ElapsedTimeS > 0 &&
-		float64(item.ExternalMovingTimeS)/float64(item.ElapsedTimeS) >= backtestMinimumMovingRatio
+	if item.GPXPath == "" || item.ExternalMovingTimeS <= 0 || item.ElapsedTimeS <= 0 {
+		return false
+	}
+	if !item.IsCategorised() {
+		return false
+	}
+	if item.RideType == ride.RideTypeIndoor || item.RideType == ride.RideTypeTransport {
+		return false
+	}
+	return true
 }
 
 func meanAbsErr(absErrors []float64) float64 {
diff --git a/cli/cli_test.go b/cli/cli_test.go
index 3e2afe2..28260e0 100644
--- a/cli/cli_test.go
+++ b/cli/cli_test.go
@@ -8,23 +8,37 @@ import (
 )
 
 func TestBacktestEligibilityExcludesShortDistances(t *testing.T) {
-	short := ride.StoredRide{GPXPath: "ride.gpx", DistanceM: 10_000 - 1, ExternalMovingTimeS: 8_000, ElapsedTimeS: 10_000}
-	minimum := ride.StoredRide{GPXPath: "ride.gpx", DistanceM: 10_000, ExternalMovingTimeS: 8_000, ElapsedTimeS: 10_000}
+	short := ride.StoredRide{GPXPath: "ride.gpx", DistanceM: 10_000 - 1, ExternalMovingTimeS: 8_000, ElapsedTimeS: 10_000, BikeType: ride.BikeTypeRoad, RideType: ride.RideTypeTraining}
+	minimum := ride.StoredRide{GPXPath: "ride.gpx", DistanceM: 10_000, ExternalMovingTimeS: 8_000, ElapsedTimeS: 10_000, BikeType: ride.BikeTypeRoad, RideType: ride.RideTypeTraining}
 
-	assert.False(t, isBacktestEligible(short))
+	assert.True(t, isBacktestEligible(short))
 	assert.True(t, isBacktestEligible(minimum))
 }
 
 func TestBacktestEligibilityExcludesLowMovingRatio(t *testing.T) {
-	low := ride.StoredRide{GPXPath: "ride.gpx", DistanceM: 10_000, ExternalMovingTimeS: 7_999, ElapsedTimeS: 10_000}
-	boundary := ride.StoredRide{GPXPath: "ride.gpx", DistanceM: 10_000, ExternalMovingTimeS: 8_000, ElapsedTimeS: 10_000}
-	noElapsed := ride.StoredRide{GPXPath: "ride.gpx", DistanceM: 10_000, ExternalMovingTimeS: 1}
+	low := ride.StoredRide{GPXPath: "ride.gpx", DistanceM: 10_000, ExternalMovingTimeS: 7_999, ElapsedTimeS: 10_000, BikeType: ride.BikeTypeRoad, RideType: ride.RideTypeTraining}
+	boundary := ride.StoredRide{GPXPath: "ride.gpx", DistanceM: 10_000, ExternalMovingTimeS: 8_000, ElapsedTimeS: 10_000, BikeType: ride.BikeTypeRoad, RideType: ride.RideTypeTraining}
+	noElapsed := ride.StoredRide{GPXPath: "ride.gpx", DistanceM: 10_000, ExternalMovingTimeS: 1, BikeType: ride.BikeTypeRoad, RideType: ride.RideTypeTraining}
 
-	assert.False(t, isBacktestEligible(low))
+	assert.True(t, isBacktestEligible(low))
 	assert.True(t, isBacktestEligible(boundary))
 	assert.False(t, isBacktestEligible(noElapsed))
 }
 
+func TestBacktestEligibilityExcludesUncategorisedAndIndoorTransport(t *testing.T) {
+	base := ride.StoredRide{GPXPath: "ride.gpx", ExternalMovingTimeS: 8_000, ElapsedTimeS: 10_000, BikeType: ride.BikeTypeRoad, RideType: ride.RideTypeTraining}
+	uncategorised := ride.StoredRide{GPXPath: "ride.gpx", ExternalMovingTimeS: 8_000, ElapsedTimeS: 10_000}
+	indoor := ride.StoredRide{GPXPath: "ride.gpx", ExternalMovingTimeS: 8_000, ElapsedTimeS: 10_000, BikeType: ride.BikeTypeRoad, RideType: ride.RideTypeIndoor}
+	transport := ride.StoredRide{GPXPath: "ride.gpx", ExternalMovingTimeS: 8_000, ElapsedTimeS: 10_000, BikeType: ride.BikeTypeRoad, RideType: ride.RideTypeTransport}
+	race := ride.StoredRide{GPXPath: "ride.gpx", ExternalMovingTimeS: 8_000, ElapsedTimeS: 10_000, BikeType: ride.BikeTypeMTB, RideType: ride.RideTypeRace}
+
+	assert.True(t, isBacktestEligible(base))
+	assert.False(t, isBacktestEligible(uncategorised))
+	assert.False(t, isBacktestEligible(indoor))
+	assert.False(t, isBacktestEligible(transport))
+	assert.True(t, isBacktestEligible(race))
+}
+
 func TestBacktestStatsUseUnroundedSamples(t *testing.T) {
 	absErrors := []float64{0, 0.1, 0.1}