def _knn_property_values(df, bounds):
clf = neighbors.KNeighborsRegressor(n_neighbors=10, weights='distance')
X = df[['LATITUDE', 'LONGITUDE']].values
y = df['PPSQFT']
clf.fit(X, y)
h = .0003 # step size in the mesh
# Plot the decision boundary. For that, we will assign a color to each
# point in the mesh [x_min, m_max]x[y_min, y_max].
yy, xx = np.meshgrid(np.arange(bounds[0][0], bounds[1][0], h),
np.arange(bounds[0][1], bounds[1][1], h))
clf.fit(X, y)
Z = clf.predict(np.c_[xx.ravel(), yy.ravel()])
# Put the result into a color plot
Z = Z.reshape(xx.shape)
MCRASPlotting.leaflet_heatmap(
yy=yy,
xx=xx,
Z=Z,
bounds=bounds,
map_path='home_value_model.html',
legend_text='Estimated value per square foot of home ($)',
description_title='Predicting home values in Boston',
description_text='Using <a target="_blank" '
'href="http://scikit-learn.org/stable/modules/generated/'
'sklearn.neighbors.KNeighborsRegressor.html#sklearn.neighbors.'
'KNeighborsRegressor">k-nearest '
'neighbors regression</a>, we fit a model over '
'the area of downtown Boston with latitude and longitude as '
'predictors and home values from 2015 assessors data to '
'create a spatial model of home prices across the city.')
def _knn_weekday_analysis(df, bounds):
X = df[['LATITUDE', 'LONGITUDE']].values
y = (1 * df['day_week'].isin(['Friday', 'Saturday', 'Sunday'])).astype(int)
neighbors_numbers = np.arange(10, 55, 1)
clf = neighbors.KNeighborsClassifier(10, weights='distance')
max_score = 0
for n_neighbors in neighbors_numbers:
clf = neighbors.KNeighborsClassifier(n_neighbors, weights='distance')
clf.fit(X, y)
if max_score + .001 > clf.score(X, y):
break
h = .0003 # step size in the mesh
# Plot the decision boundary. For that, we will assign a color to each
# point in the mesh [x_min, m_max]x[y_min, y_max].
yy, xx = np.meshgrid(np.arange(bounds[0][0], bounds[1][0], h),
np.arange(bounds[0][1], bounds[1][1], h))
Z = clf.predict(np.c_[xx.ravel(), yy.ravel()])
# Put the result into a color plot
Z = Z.reshape(xx.shape)
MCRASPlotting.leaflet_heatmap(yy=yy, xx=xx, Z=Z, bounds=bounds, map_path='crime_knn_weekday.html',
legend_text='Weekday (0) vs Weekend (1)', tiles='osm_mapnik',
description_title='Predicting weekend vs weekday crimes in Boston',
description_text='Using <a target="_blank" '
'href="http://scikit-learn.org/stable/modules/generated'
'/sklearn.neighbors.KNeighborsClassifier.html">k-nearest '
'neighbors classification</a>, we fit a model over '
'the area of downtown Boston with latitude and longitude as '
'predictors and all crimes from January to August 2015, '
'flagged with day of the week they were committed.')
def _knn_property_values(df, bounds):
clf = neighbors.KNeighborsRegressor(n_neighbors=10, weights='distance')
X = df[['LATITUDE', 'LONGITUDE']].values
y = df['PPSQFT']
clf.fit(X, y)
h = .0003 # step size in the mesh
# Plot the decision boundary. For that, we will assign a color to each
# point in the mesh [x_min, m_max]x[y_min, y_max].
yy, xx = np.meshgrid(np.arange(bounds[0][0], bounds[1][0], h),
np.arange(bounds[0][1], bounds[1][1], h))
clf.fit(X, y)
Z = clf.predict(np.c_[xx.ravel(), yy.ravel()])
# Put the result into a color plot
Z = Z.reshape(xx.shape)
MCRASPlotting.leaflet_heatmap(yy=yy, xx=xx, Z=Z, bounds=bounds, map_path='home_value_model.html',
legend_text='Estimated value per square foot of home ($)',
description_title='Predicting home values in Boston',
description_text='Using <a target="_blank" '
'href="http://scikit-learn.org/stable/modules/generated/'
'sklearn.neighbors.KNeighborsRegressor.html#sklearn.neighbors.'
'KNeighborsRegressor">k-nearest '
'neighbors regression</a>, we fit a model over '
'the area of downtown Boston with latitude and longitude as '
'predictors and home values from 2015 assessors data to '
'create a spatial model of home prices across the city.')