Table 2 Summary of models performances according to intercept and slope coefficients.

From: Using global navigation satellite systems for modeling athletic performances in elite football players

Model

Target

Population

Aggregation

\(\overline{MAPE}\)

Multi-modal Ensemble

Intercept

I

N/A

0.076

LSTM (raw)

Intercept

G

N/A

0.077

Ridge

Intercept

G

Exponential

0.080

Ridge

Intercept

G

Mean

0.080

Uni-modal Ensemble

Intercept

I

N/A

0.080

LSTM

Intercept

G

Exponential

0.084

LSTM

Intercept

G

Mean

0.084

Ridge

Intercept

I

Mean

0.085

Ridge

Intercept

I

Exponential

0.085

LSTM (raw)

Intercept

I

N/A

0.088

LSTM

Intercept

I

Mean

0.089

LSTM

Intercept

I

Exponential

0.090

LSTM

Slope

G

Mean

0.114

LSTM

Slope

G

Exponential

0.114

Uni-modal Ensemble

Slope

I

N/A

0.115

Multi-modal Ensemble

Slope

I

N/A

0.116

RIDGE

Slope

G

Mean

0.116

RIDGE

Slope

G

Exponential

0.116

LSTM (raw)

Slope

G

N/A

0.119

LSTM (raw)

Slope

I

N/A

0.121

LSTM

Slope

I

Mean

0.126

RIDGE

Slope

I

Mean

0.128

LSTM

Slope

I

Exponential

0.128

RIDGE

Slope

I

Exponential

0.129

  1. \(\overline{MAPE}\) represents the averaged MAPE over individuals and validation folders. The population represents either models computed over the group of players (G) or individually computed models (I).
  2. Significant values are in [bold].
Back to article page