@lgodard/outliers-interquartile-range NPM

Outliers Interquartile Range

Suggests outliers based on Interquartile Range method

The calculated bounds are

lower_bound = q25 - k * interquartile_range
upper_bound = q75 + k * interquartile_range

Any value outside these bounds is suggested as an outlier.

see https://en.wikipedia.org/wiki/Interquartile_range#Outliers for details

Installation

npm install @lgodard/outliers-interquartile-range

Example

const outliersEngine = require('outliers-interquartile-range');

const options = {sorted: false, k: 1.5};

// array_values is 1-D array to be analyzed
const results = outliersEngine.getOutliers(array_values, options);

Documentation

Option	type	default	description
`sorted`	`boolean`	`false`	The submited `data_array` is already sorted, avoiding costly operation
`k`	`number`	`1.5`	Interquartile range multiplier defining the threshold

Results

The getOutliers method returns an object with the following entries

Entry	description
`outliers`	`Object` `upper` & `lower` outliers - see above
`stats`	`Object` calculated values `q25`, `q75` & `iqr` (interquartile range)
`parameters`	`Object` the used parameters combining submitted options and defaults

The upper & lower objects contain

Entry	description
`threshold`	`number` The limit for the value being suggested as outlier
`values`	`Array` The suggested outlier values
`indexes`	`Array` The indexes against submitted `data_array` suggested as outliers

results = { 
    'outliers': {
        'upper': {
            'threshold': 64540.095,
            'values': [ 193568.04, 128104.71, 235793.39, 157432.6 ],
            'indexes': [ 11, 36, 43, 82]
        },
        'lower': {
            'threshold': -36329.865,
            'values': [],
            'indexes': []
        }
    },
    'parameters': {
        'sorted': false,
        'k': 1.5
    },
    'stats': {
        'q25': 1496.37,
        'q75': 26713.86,
        'iqr': 25217.49
    }
};

Acknowledgements

The main work (quartiles calculation) is done through the summary package. Many thanks to the author.

License

MIT

summary

@everything-registry/sub-chunk-545

1.0.3

3 years ago