Skip to main content

Table 4 Trained accuracy of the LSTM and GRU architectural models

From: Implementation of Long Short-Term Memory and Gated Recurrent Units on grouped time-series data to predict stock prices accurately

Company

Model

Upper accucacy

(MAPE)

Average accuracy

(RMSPE)

Lower accuracy

(RMDPE)

LSTM

GRU

LSTM

GRU

LSTM

GRU

AMZN

Model 1

96.78%

94.39%

95.79%

93.51%

83.40%

80.89%

Model 2

94.28%

92.74%

92.85%

91.19%

78.31%

74.65%

Model 3

95.36%

92.92%

94.30%

90.44%

82.80%

66.64%

Model 4

95.75%

95.75%

94.52%

94.52%

82.56%

82.56%

GOOGL

Model 1

96.14%

96.64%

94.77%

95.62%

80.65%

85.81%

Model 2

95.61%

97.58%

94.35%

96.97%

82.22%

84.70%

Model 3

96.99%

97.13%

95.99%

96.28%

84.00%

85.16%

Model 4

94.68%

94.68%

93.71%

93.71%

83.47%

83.47%

BLL

Model 1

97.00%

98.48%

96.22%

97.98%

87.32%

90.73%

Model 2

96.62%

97.15%

95.11%

96.56%

82.34%

87.32%

Model 3

96.85%

96.11%

95.52%

95.03%

83.59%

83.58%

Model 4

96.24%

96.24%

95.71%

95.71%

86.74%

86.74%

QCOM

Model 1

98.00%

97.85%

97.04%

97.06%

78.96%

82.98%

Model 2

97.66%

96.23%

96.56%

95.75%

75.94%

81.57%

Model 3

97.55%

97.09%

96.41%

96.23%

76.70%

79.70%

Model 4

97.95%

97.95%

97.27%

97.27%

83.11%

83.11%