joaofig

#
# Generates the simulated server log
#
import csv
import numpy as np
import random

base = 750
amplitude = 500
growth = 1e-6

joaofig

using System;

namespace Sagaceco.TimeSeries.Patterns.Models
{
    public class ExponentialMovingModel
    {
        private double average = 0.0;
        private double variance = 0.0;

        public ExponentialMovingModel()

joaofig

namespace Sagaceco.ServerLog
{
    public class WeeklyLogModel
    {
        private ExponentialMovingModel[] models = new ExponentialMovingModel[2016];
        private double radius = 3.0;

        public WeeklyLogModel()
        {
            for(int i = 0; i < models.Length; i++)

joaofig

public void Update(double x, double y)
{
    count++;
    sumX    += x;
    sumY    += y;
    sumXY   += x * y;
    sumXX   += x * x;
    sumYY   += y * y;

    double meanX    = sumX / count;

joaofig

public bool IsOutlier(double radius, double x, double y)
{
    double yHat = GetValue( x );
    double studentizedResidue = Math.Abs(y - yHat) / Math.Sqrt( GetResidualVariance() );

    return studentizedResidue > radius;
}

private double GetResidualVariance()
{

joaofig

# Parameters
eps_in_meters = 100.0
num_samples = 10

joaofig

# Cluster the data
from sklearn.cluster import DBSCAN

earth_perimeter = 40070000.0  # In meters
eps_in_radians = eps_in_meters / earth_perimeter * (2 * math.pi)

uk_acc['cluster'] = DBSCAN(eps=eps_in_radians, min_samples=num_samples, metric='haversine').fit_predict(uk_acc[['rad_lat', 'rad_lng']])

joaofig

# Group the observations by cluster identifier
groups = uk_acc.groupby('cluster')

joaofig

# Create the list of cluster blobs
from shapely.ops import cascaded_union

clusters = list()
blobs = list()
counts = list()

for cluster_id, points in groups:
    if cluster_id >= 0:
        buffer_radius = eps_in_meters * 0.6

joaofig

# Create the GeoDataFrame from the cluster numbers and blobs
data = { 'cluster': clusters, 'polygon': blobs, 'count': counts }

cluster_gdf = gpd.GeoDataFrame(pd.DataFrame(data), geometry='polygon')
cluster_gdf.crs = {'init': 'epsg:4326'}

ax = cluster_gdf.geometry.plot(linewidth=2.0, color='red', edgecolor='red')