Figure 8
The figure shows four bar graphs arranged in a two-by-two grid. The top graph is titled “Prophet underscore Error Distribution.” The horizontal axis is labeled “Error” and ranges from negative 60000 to negative 10000 in increments of 10000 units. The vertical axis is labeled “Frequency” and ranges from 0 to 4 in increments of 1 unit. The histogram contains multiple bars showing error frequencies, with higher bar counts concentrated between negative 30000 and negative 10000. The tallest bar at negative 30000 and between negative 20000 to negative 10000 has a height of 4. The second graph on the top right is titled “A R I M A underscore Error Distribution.” The horizontal axis is labeled “Error” and ranges from negative 50000 to 10000 in increments of 10000 units. The vertical axis is labeled “Frequency” and ranges from 0 to 4 in increments of 1 unit. The histogram bars show most frequencies between negative 30000 and negative 7000, with a single outlier bar near 8000. The tallest bar at negative 5000 has a height of 4. The third graph on the bottom left is titled “X G Boost underscore Error Distribution.” The horizontal axis is labeled “Error” and ranges from negative 50000 to 10000 in increments of 10000 units. The vertical axis is labeled “Frequency” and ranges from 0 to 5 in increments of 1 unit. The histogram bars show most frequencies between negative 25000 and 10000, with a single bar near negative 50000. The tallest bar at negative 25000 has a height of 5. The fourth graph on the bottom right is titled “L S T M underscore Error Distribution.” The horizontal axis is labeled “Error” and ranges from negative 20000 to 10000 in increments of 10000 units. The vertical axis is labeled “Frequency” and ranges from 0 to 4 in increments of 1 unit. The histogram bars show most frequencies between negative 15000 and 10000, with one bar near positive 12000. The tallest bar near 0 has a height of 4. Note: All numerical data values are approximated.Distribution of forecast errors across models: Prophet, ARIMA, XGBoost and LSTM