Forecasting Data

[srinivasyl]

I've a data set of 120 items i need to forecast next item or next 5 items how to do it.

I've tried below example
`var forecastDatapoint = 121;

var coeffs = t.ARMaxEntropy({
    data:	t.data.slice(0,120)
});
console.log(coeffs);
 

var forecast	= 0;	// Init the value at 0.
for (var i=0;i<coeffs.length;i++) {	// Loop through the coefficients
    forecast -= t.data[10-i][1]*coeffs[i];
}
console.log("forecast",forecast);`

I have noticed that you've no where used forecastDatapoint this variable in above example so how exactly you're calculating 121 (11 in example) data point
also tired the sliding_regression_forecast but it failed to show red dot indicate at which point the forecast starts.
can you help in forecasting next 5 data points?

[26medias]

// Calculate the AR coefficients of the 120 previous points
var coeffs = t.ARMaxEntropy({
    data:   t.data.slice(0,120) // regression_forecast_optimize() give you the number of sample to use
});
// You might want to use only the last 20 points maybe to have a forecast that considers only the latest movements and not the entire history? It'd work a lot better for real-life data:
var coeffs = t.ARMaxEntropy({
    data:   t.data.slice(t.data.length-20)
});

// Output the coefficients to the console
console.log(coeffs);

// Calculate the forecasted value of the 21st datapoint using the AR coefficients:
var forecast    = 0;    // Init the value at 0.
for (var i=0;i<coeffs.length;i++) { // Loop through the coefficients
    forecast -= t.data[20-i][1]*coeffs[i];
    // Explanation for that line:
    // t.data contains the current dataset, which is in the format [ [date, value], [date,value], ... ]
    // For each coefficient, we substract from "forecast" the value of the "N - x" datapoint's value, multiplicated by the coefficient, where N is the last known datapoint value, and x is the coefficient's index.
}

console.log("forecast",forecast);

// Now push that number to t.data and do the same thing again:
// Get the last 20 points (so the 19 latest of the point from the previous data + the forecast we just found making the 20th point)
t.data.push([new Date(), forecast]);
var coeffs = t.ARMaxEntropy({
    data:   t.data.slice(t.data.length-20)
});
//...

If you look at the code of , this is what it does:

• Find the coefficient: https://github.com/26medias/timeseries-analysis/blob/master/timeseries-analysis.js#L795
• Add the forecast to the data: https://github.com/26medias/timeseries-analysis/blob/master/timeseries-analysis.js#L799
• Loop again until you have obtained the number of forecasts: https://github.com/26medias/timeseries-analysis/blob/master/timeseries-analysis.js#L790

If you use sliding_regression_forecast(), make sure that your sample size if only a few points, not the entire size of your data.

[dushyantbangal]

@26medias if I am collecting temperature data every 10 minutes. So I'll have 144 records for the whole day.
If I need to forecast the next day's data, won't I need to use the whole dataset in order for it to understand the pattern?

[26medias]

In that case, yes, it makes sense

If it still doesn't work, can you send a demo of your code on https://plnkr.co/ or another JS editor?