def test_vega_heat_map():
vega = vega_heatmap(1900, 1410,
[-73.998427, 40.730309, -73.954348, 40.780816], 10.0,
"EPSG:3857").build()
vega_dict = json.loads(vega)
assert vega_dict["width"] == 1900
assert vega_dict["height"] == 1410
assert vega_dict["marks"][0]["encode"]["enter"]["bounding_box"]["value"][
0] == -73.998427
assert vega_dict["marks"][0]["encode"]["enter"]["bounding_box"]["value"][
1] == 40.730309
assert vega_dict["marks"][0]["encode"]["enter"]["bounding_box"]["value"][
2] == -73.954348
assert vega_dict["marks"][0]["encode"]["enter"]["bounding_box"]["value"][
3] == 40.780816
assert vega_dict["marks"][0]["encode"]["enter"]["map_zoom_level"][
"value"] == 10.0
assert vega_dict["marks"][0]["encode"]["enter"]["coordinate_system"][
"value"] == "EPSG:3857"
assert vega_dict["marks"][0]["encode"]["enter"]["aggregation_type"][
"value"] == "sum"
vega = vega_heatmap(1900, 1410,
[-73.998427, 40.730309, -73.954348, 40.780816],
10.0).build()
vega_dict = json.loads(vega)
assert vega_dict["marks"][0]["encode"]["enter"]["coordinate_system"][
"value"] == "EPSG:4326"
def draw_heat_map(spark):
start_time = time.time()
df = spark.read.format("csv").option("header", True).option(
"delimiter", ","
).schema(
"VendorID string, tpep_pickup_datetime timestamp, tpep_dropoff_datetime timestamp, passenger_count long, trip_distance double, pickup_longitude double, pickup_latitude double, dropoff_longitude double, dropoff_latitude double, fare_amount double, tip_amount double, total_amount double, buildingid_pickup long, buildingid_dropoff long, buildingtext_pickup string, buildingtext_dropoff string"
).load("file:///tmp/0_5M_nyc_taxi_and_building.csv").cache()
df.createOrReplaceTempView("nyc_taxi")
register_funcs(spark)
res = spark.sql(
"select ST_Point(pickup_longitude, pickup_latitude) as point, passenger_count as w from nyc_taxi where ST_Within(ST_Point(pickup_longitude, pickup_latitude), ST_GeomFromText('POLYGON ((-73.998427 40.730309, -73.954348 40.730309, -73.954348 40.780816 ,-73.998427 40.780816, -73.998427 40.730309))'))"
)
res.show()
vega = vega_heatmap(1024, 896,
[-73.998427, 40.730309, -73.954348, 40.780816], 10.0,
'EPSG:4326')
res = heatmap(vega, res)
save_png(res, '/tmp/heatmap.png')
spark.sql("show tables").show()
spark.catalog.dropGlobalTempView("nyc_taxi")
print("--- %s seconds ---" % (time.time() - start_time))
def test_heat_map():
x_data = []
y_data = []
c_data = []
x_data.append(-73.96524)
x_data.append(-73.96118)
x_data.append(-73.97324)
x_data.append(-73.98456)
y_data.append(40.73747)
y_data.append(40.74507)
y_data.append(40.75890)
y_data.append(40.77654)
c_data.append(10)
c_data.append(20)
c_data.append(30)
c_data.append(40)
arr_x = pandas.Series(x_data)
arr_y = pandas.Series(y_data)
arr_c = pandas.Series(c_data)
points = arctern.ST_Point(arr_x, arr_y)
vega = vega_heatmap(
1024,
896,
bounding_box=[-73.998427, 40.730309, -73.954348, 40.780816],
map_zoom_level=13.0,
coordinate_system='EPSG:4326')
heat_map1 = arctern.heat_map_layer(vega, points, arr_c)
save_png(heat_map1, "/tmp/test_heat_map1.png")
def heatmap(ax,
points,
weights,
bounding_box,
map_zoom_level=None,
coordinate_system='EPSG:3857',
aggregation_type='max',
**extra_contextily_params):
"""
:type ax: AxesSubplot
:param ax: Matplotlib axes object on which to add the basemap.
:type points: Series(dtype: object)
:param points: Points in WKB form
:type bounding_box: (float, float, float, float)
:param bounding_box: The bounding rectangle, as a [left, upper, right, lower]-tuple.
value should be of :coordinate_system:
:type coordinate_system: str
:param coordinate_system: either 'EPSG:4326' or 'EPSG:3857'
:type extra_contextily_params: dict
:param extra_contextily_params: extra parameters for contextily.add_basemap.
See https://contextily.readthedocs.io/en/latest/reference.html
"""
from matplotlib import pyplot as plt
import contextily as cx
bbox = _transform_bbox(bounding_box, coordinate_system, 'epsg:3857')
w, h = _get_recom_size(bbox[2] - bbox[0], bbox[3] - bbox[1])
if map_zoom_level is None:
map_zoom_level = _calc_zoom(bounding_box, coordinate_system)
vega = vega_heatmap(w,
h,
bounding_box=bounding_box,
map_zoom_level=map_zoom_level,
aggregation_type=aggregation_type,
coordinate_system=coordinate_system)
hexstr = arctern.heat_map_layer(vega, points, weights)
f = io.BytesIO(base64.b64decode(hexstr))
img = plt.imread(f)
ax.set(xlim=(bbox[0], bbox[2]), ylim=(bbox[1], bbox[3]))
cx.add_basemap(ax, **extra_contextily_params)
ax.imshow(img,
alpha=img[:, :, 3],
extent=(bbox[0], bbox[2], bbox[1], bbox[3]))
def test_heat_map():
x_data = []
y_data = []
c_data = []
for i in range(0, 5):
x_data.append(i + 50)
y_data.append(i + 50)
c_data.append(i + 50)
arr_x = pandas.Series(x_data)
arr_y = pandas.Series(y_data)
points = arctern.ST_Point(arr_x, arr_y)
arr_c = pandas.Series(y_data)
vega = vega_heatmap(1024, 896, [-73.998427, 40.730309, -73.954348, 40.780816], 10.0, 'EPSG:4326')
heat_map1 = arctern.heat_map(vega, points, arr_c)
save_png(heat_map1, "/tmp/test_heat_map1.png")
def draw_heat_map(spark):
df = spark.read.format("csv").option("header", True).option(
"delimiter", ","
).schema(
"VendorID string, tpep_pickup_datetime timestamp, tpep_dropoff_datetime timestamp, passenger_count long, trip_distance double, pickup_longitude double, pickup_latitude double, dropoff_longitude double, dropoff_latitude double, fare_amount double, tip_amount double, total_amount double, buildingid_pickup long, buildingid_dropoff long, buildingtext_pickup string, buildingtext_dropoff string"
).load(data_path).cache()
df.show(20, False)
df.createOrReplaceTempView("nyc_taxi")
register_funcs(spark)
res = spark.sql(
"select ST_Point(pickup_longitude, pickup_latitude) as point, passenger_count as w from nyc_taxi where ST_Within(ST_Point(pickup_longitude, pickup_latitude), 'POLYGON ((-73.998427 40.730309, -73.954348 40.730309, -73.954348 40.780816 ,-73.998427 40.780816, -73.998427 40.730309))')"
)
vega = vega_heatmap(1024, 896, 10.0,
[-73.998427, 40.730309, -73.954348, 40.780816],
'EPSG:4326')
res = heatmap(res, vega)
save_png(res, '/tmp/heatmap.png')
spark.sql("show tables").show()
spark.catalog.dropGlobalTempView("nyc_taxi")
def test_heat_map():
x_data = []
y_data = []
c_data = []
for i in range(0, 5):
x_data.append(i + 50)
y_data.append(i + 50)
c_data.append(i + 50)
arr_x = pandas.Series(x_data)
arr_y = pandas.Series(y_data)
arr_c = pandas.Series(y_data)
vega_heat_map = vega_heatmap(
1024, 896, 10.0, [-73.998427, 40.730309, -73.954348, 40.780816],
'EPSG:4326')
vega_json = vega_heat_map.build()
heat_map1 = arctern.heat_map(arr_x, arr_y, arr_c,
vega_json.encode('utf-8'))
save_png(heat_map1, "/tmp/test_heat_map1.png")
def heatmap(ax,
points,
weights,
bounding_box,
map_zoom_level=None,
coordinate_system='EPSG:3857',
aggregation_type='max',
**extra_contextily_params):
"""
Plot heatmap in matplotlibs
:type ax: AxesSubplot
:param ax: Matplotlib axes object on which to add the basemap.
:type points: GeoSeries
:param points: Sequence of Points
:type weights: Series(dtype: float|int64)
:param weights: Weights of point intensity
:type bounding_box: list
:param bounding_box: Specify the bounding rectangle [west, south, east, north].
:type map_zoom_level: int
:param map_zoom_level: Zoom level of heatmap. Default as auto.
:type coordinate_system: str
:param coordinate_system: Coordinate Reference System of the geometry objects.
Must be SRID formed, e.g. 'EPSG:4326' or 'EPSG:3857'
Default as 'EPSG:3857'
:type aggregation_type: str
:param aggregation_type: Aggregation type of data processing. Default as 'max'
:type extra_contextily_params: dict
:param extra_contextily_params: Extra parameters will be passed to contextily.add_basemap.
See https://contextily.readthedocs.io/en/latest/reference.html for details
:example:
>>> import pandas as pd
>>> import numpy as np
>>> import arctern
>>> import matplotlib.pyplot as plt
>>> # read from test_data.csv
>>> # Download link: https://raw.githubusercontent.com/arctern-io/arctern-resources/benchmarks/benchmarks/dataset/layer_rendering_test_data/test_data.csv
>>> df = pd.read_csv("/path/to/test_data.csv", dtype={'longitude':np.float64, 'latitude':np.float64, 'color_weights':np.float64, 'size_weights':np.float64, 'region_boundaries':np.object})
>>> points = arctern.GeoSeries.point(df['longitude'], df['latitude'])
>>>
>>> # plot heatmap
>>> fig, ax = plt.subplots(figsize=(10, 6), dpi=200)
>>> arctern.plot.heatmap(ax, points, df['color_weights'], bounding_box=[-74.01424568752932, 40.72759334104623, -73.96056823889673, 40.76721122683304], coordinate_system='EPSG:4326')
>>> plt.show()
"""
from matplotlib import pyplot as plt
import contextily as cx
bbox = _transform_bbox(bounding_box, coordinate_system, 'epsg:3857')
w, h = _get_recom_size(bbox[2] - bbox[0], bbox[3] - bbox[1])
if map_zoom_level is None:
map_zoom_level = _calc_zoom(bounding_box, coordinate_system)
vega = vega_heatmap(w,
h,
bounding_box=bounding_box,
map_zoom_level=map_zoom_level,
aggregation_type=aggregation_type,
coordinate_system=coordinate_system)
hexstr = arctern.heat_map_layer(vega, points, weights)
f = io.BytesIO(base64.b64decode(hexstr))
img = plt.imread(f)
ax.set(xlim=(bbox[0], bbox[2]), ylim=(bbox[1], bbox[3]))
cx.add_basemap(ax, **extra_contextily_params)
ax.imshow(img,
alpha=img[:, :, 3],
extent=(bbox[0], bbox[2], bbox[1], bbox[3]))
ax.axis('off')
def db_query():
"""
/db/query handler
"""
log.INSTANCE.info('POST /db/query: {}'.format(request.json))
if not utils.check_json(request.json, 'id') \
or not utils.check_json(request.json, 'query') \
or not utils.check_json(request.json['query'], 'type') \
or not utils.check_json(request.json['query'], 'sql'):
return jsonify(status='error', code=-1, message='query format error')
query_sql = request.json['query']['sql']
query_type = request.json['query']['type']
content = {}
content['sql'] = query_sql
content['err'] = False
db_instance = db.CENTER.get(str(request.json['id']), None)
if db_instance is None:
return jsonify(status="error",
code=-1,
message='there is no database whose id equal to ' +
str(request.json['id']))
if query_type == 'sql':
res = db_instance.run_for_json(query_sql)
data = []
for row in res:
obj = json.loads(row)
data.append(obj)
content['result'] = data
else:
if not utils.check_json(request.json['query'], 'params'):
return jsonify(status='error',
code=-1,
message='query format error')
query_params = request.json['query']['params']
res = db_instance.run(query_sql)
if query_type == 'point':
vega = vega_pointmap(int(query_params['width']),
int(query_params['height']),
query_params['point']['bounding_box'],
int(query_params['point']['point_size']),
query_params['point']['point_color'],
float(query_params['point']['opacity']),
query_params['point']['coordinate_system'])
data = pointmap(vega, res)
content['result'] = data
elif query_type == 'heat':
vega = vega_heatmap(int(query_params['width']),
int(query_params['height']),
query_params['heat']['bounding_box'],
float(query_params['heat']['map_zoom_level']),
query_params['heat']['coordinate_system'],
query_params['heat']['aggregation_type'])
data = heatmap(vega, res)
content['result'] = data
elif query_type == 'choropleth':
vega = vega_choroplethmap(
int(query_params['width']), int(query_params['height']),
query_params['choropleth']['bounding_box'],
query_params['choropleth']['color_gradient'],
query_params['choropleth']['color_bound'],
float(query_params['choropleth']['opacity']),
query_params['choropleth']['coordinate_system'],
query_params['choropleth']['aggregation_type'])
data = choroplethmap(vega, res)
content['result'] = data
elif query_type == 'weighted':
vega = vega_weighted_pointmap(
int(query_params['width']), int(query_params['height']),
query_params['weighted']['bounding_box'],
query_params['weighted']['color_gradient'],
query_params['weighted']['color_bound'],
query_params['weighted']['size_bound'],
float(query_params['weighted']['opacity']),
query_params['weighted']['coordinate_system'])
data = weighted_pointmap(vega, res)
content['result'] = data
elif query_type == 'icon':
vega = vega_icon(int(query_params['width']),
int(query_params['height']),
query_params['icon']['bounding_box'],
query_params['icon']['icon_path'],
query_params['icon']['coordinate_system'])
data = icon_viz(vega, res)
content['result'] = data
else:
return jsonify(status="error",
code=-1,
message='{} not support'.format(query_type))
return jsonify(status="success", code=200, data=content)
bounding_box=[pos1[0], pos1[1], pos2[0], pos2[1]],
color_gradient=["#115f9a", "#d0f400"],
color_bound=[1, 50],
size_bound=[3, 15],
opacity=1.0,
coordinate_system="EPSG:4326")
png = weighted_point_map_layer(vega,
ST_Point(pickup_df.pickup_longitude,
pickup_df.pickup_latitude),
color_weights=df.head(limit_num).fare_amount,
size_weights=df.head(limit_num).total_amount)
save_png(png, "/tmp/arctern_weighted_pointmap_pandas.png")
vega = vega_heatmap(1024,
384,
bounding_box=[pos1[0], pos1[1], pos2[0], pos2[1]],
map_zoom_level=13.0,
coordinate_system="EPSG:4326")
png = heat_map_layer(
vega, ST_Point(pickup_df.pickup_longitude, pickup_df.pickup_latitude),
df.head(limit_num).fare_amount)
save_png(png, "/tmp/arctern_heatmap_pandas.png")
vega = vega_choroplethmap(1024,
384,
bounding_box=[pos1[0], pos1[1], pos2[0], pos2[1]],
color_gradient=["#115f9a", "#d0f400"],
color_bound=[2.5, 5],
opacity=1.0,
coordinate_system="EPSG:4326")
png = choropleth_map_layer(vega,
def heatmap(ax,
points,
weights,
bounding_box,
map_zoom_level=None,
coordinate_system='EPSG:3857',
aggregation_type='max',
**extra_contextily_params):
"""
Plots a heat map in matplotlib.
Parameters
----------
ax : matplotlib.axes.Axes
Axes where geometries will be plotted.
points : GeoSeries
Sequence of points.
weights : Series
Weights of point intensity.
bounding_box : list
Bounding box of the map. For example, [west, south, east, north].
map_zoom_level : [type], optional
Zoom level of the map by default 'auto'.
coordinate_system : str, optional
The Coordinate Reference System (CRS) set to all geometries, by default 'EPSG:3857'.
Only supports SRID as a WKT representation of CRS by now.
aggregation_type : str, optional
Aggregation type, by default 'max'.
**extra_contextily_params: dict
Extra parameters passed to `contextily.add_basemap. <https://contextily.readthedocs.io/en/latest/reference.html>`_
Examples
-------
>>> import pandas as pd
>>> import numpy as np
>>> import arctern
>>> import matplotlib.pyplot as plt
>>> # read from test_data.csv
>>> # Download link: https://raw.githubusercontent.com/arctern-io/arctern-resources/benchmarks/benchmarks/dataset/layer_rendering_test_data/test_data.csv
>>> df = pd.read_csv("/path/to/test_data.csv", dtype={'longitude':np.float64, 'latitude':np.float64, 'color_weights':np.float64, 'size_weights':np.float64, 'region_boundaries':np.object})
>>> points = arctern.GeoSeries.point(df['longitude'], df['latitude'])
>>>
>>> # plot heatmap
>>> fig, ax = plt.subplots(figsize=(10, 6), dpi=200)
>>> arctern.plot.heatmap(ax, points, df['color_weights'], bounding_box=[-74.01424568752932, 40.72759334104623, -73.96056823889673, 40.76721122683304], coordinate_system='EPSG:4326')
>>> plt.show()
"""
from matplotlib import pyplot as plt
import contextily as cx
bbox = _transform_bbox(bounding_box, coordinate_system, 'epsg:3857')
w, h = _get_recom_size(bbox[2] - bbox[0], bbox[3] - bbox[1])
if map_zoom_level is None:
map_zoom_level = _calc_zoom(bounding_box, coordinate_system)
vega = vega_heatmap(w,
h,
bounding_box=bounding_box,
map_zoom_level=map_zoom_level,
aggregation_type=aggregation_type,
coordinate_system=coordinate_system)
hexstr = arctern.heat_map_layer(vega, points, weights)
f = io.BytesIO(base64.b64decode(hexstr))
img = plt.imread(f)
ax.set(xlim=(bbox[0], bbox[2]), ylim=(bbox[1], bbox[3]))
cx.add_basemap(ax, **extra_contextily_params)
ax.imshow(img,
alpha=img[:, :, 3],
extent=(bbox[0], bbox[2], bbox[1], bbox[3]))
ax.axis('off')
def db_query():
"""
/db/query handler
"""
if not utils.check_json(request.json, 'id') \
or not utils.check_json(request.json, 'query') \
or not utils.check_json(request.json['query'], 'type') \
or not utils.check_json(request.json['query'], 'sql'):
return jsonify(status='error', code=-1, message='query format error')
query_sql = request.json['query']['sql']
query_type = request.json['query']['type']
content = {}
content['sql'] = query_sql
content['err'] = False
if query_type == 'sql':
res = spark.Spark.run_for_json(query_sql)
data = []
for row in res:
obj = json.loads(row)
data.append(obj)
content['result'] = data
else:
if not utils.check_json(request.json['query'], 'params'):
return jsonify(status='error',
code=-1,
message='query format error')
query_params = request.json['query']['params']
res = spark.Spark.run(query_sql)
if query_type == 'point':
vega = vega_pointmap(int(query_params['width']),
int(query_params['height']),
query_params['point']['bounding_box'],
int(query_params['point']['stroke_width']),
query_params['point']['stroke'],
float(query_params['point']['opacity']),
query_params['point']['coordinate'])
data = pointmap(res, vega)
content['result'] = data
elif query_type == 'heat':
vega = vega_heatmap(int(query_params['width']),
int(query_params['height']),
float(query_params['heat']['map_scale']),
query_params['heat']['bounding_box'],
query_params['heat']['coordinate'])
data = heatmap(res, vega)
content['result'] = data
elif query_type == 'choropleth':
vega = vega_choroplethmap(
int(query_params['width']), int(query_params['height']),
query_params['choropleth']['bounding_box'],
query_params['choropleth']['color_style'],
query_params['choropleth']['rule'],
float(query_params['choropleth']['opacity']),
query_params['choropleth']['coordinate'])
data = choroplethmap(res, vega)
content['result'] = data
else:
return jsonify(status="error",
code=-1,
message='{} not support'.format(query_type))
return jsonify(status="success", code=200, data=content)
def run_test_heat_map(spark):
df = spark.read.format("csv").option("header", True).option("delimiter", ",").schema(
"VendorID string, tpep_pickup_datetime timestamp, tpep_dropoff_datetime timestamp, passenger_count long, "
"trip_distance double, pickup_longitude double, pickup_latitude double, dropoff_longitude double, "
"dropoff_latitude double, fare_amount double, tip_amount double, total_amount double, buildingid_pickup long, "
"buildingid_dropoff long, buildingtext_pickup string, buildingtext_dropoff string").load(
file_path).cache()
df.createOrReplaceTempView("nyc_taxi")
register_funcs(spark)
res = spark.sql(
"select ST_Point(pickup_longitude, pickup_latitude) as point, passenger_count as w from nyc_taxi where ST_Within(ST_Point(pickup_longitude, pickup_latitude), 'POLYGON ((-73.998427 40.730309, -73.954348 40.730309, -73.954348 40.780816 ,-73.998427 40.780816, -73.998427 40.730309))')")
# 1 size:1024*896, map_scale: 10.0
vega_1 = vega_heatmap(1024, 896, 10.0, [-73.998427, 40.730309, -73.954348, 40.780816], 'EPSG:4326')
baseline1 = heatmap(res, vega_1)
heat_map1_1 = heatmap(res, vega_1)
heat_map1_2 = heatmap(res, vega_1)
baseline_png1 = png_path + "heat_map_nyc_1.png"
save_png(baseline1, baseline_png1)
save_png(heat_map1_1, png_path + "test_heat_map_nyc_1-1.png")
save_png(heat_map1_2, png_path + "test_heat_map_nyc_1-2.png")
# 2 map_scale: 0.0
vega_2 = vega_heatmap(1024, 896, 0.0, [-73.998427, 40.730309, -73.954348, 40.780816], 'EPSG:4326')
baseline2 = heatmap(res, vega_2)
heat_map2_1 = heatmap(res, vega_2)
heat_map2_2 = heatmap(res, vega_2)
baseline_png2 = png_path + "heat_map_nyc_2.png"
save_png(baseline2, baseline_png2)
save_png(heat_map2_1, png_path + "test_heat_map_nyc_2-1.png")
save_png(heat_map2_2, png_path + "test_heat_map_nyc_2-2.png")
# 3 map_scale: 12.0
vega_3 = vega_heatmap(1024, 896, 12.0, [-73.998427, 40.730309, -73.954348, 40.780816], 'EPSG:4326')
baseline3 = heatmap(res, vega_3)
heat_map3_1 = heatmap(res, vega_3)
heat_map3_2 = heatmap(res, vega_3)
baseline_png3 = png_path + "heat_map_nyc_3.png"
save_png(baseline3, baseline_png3)
save_png(heat_map3_1, png_path + "test_heat_map_nyc_3-1.png")
save_png(heat_map3_2, png_path + "test_heat_map_nyc_3-2.png")
# 4 map_scale: 5.5
vega_4 = vega_heatmap(1024, 896, 5.5, [-73.998427, 40.730309, -73.954348, 40.780816], 'EPSG:4326')
baseline4 = heatmap(res, vega_4)
heat_map4_1 = heatmap(res, vega_4)
heat_map4_2 = heatmap(res, vega_4)
baseline_png4 = png_path + "heat_map_nyc_4.png"
save_png(baseline4, baseline_png4)
save_png(heat_map4_1, png_path + "test_heat_map_nyc_4-1.png")
save_png(heat_map4_2, png_path + "test_heat_map_nyc_4-2.png")
# 5 size:200*200
vega_5 = vega_heatmap(200, 200, 10.0, [-73.998427, 40.730309, -73.954348, 40.780816], 'EPSG:4326')
baseline5 = heatmap(res, vega_5)
heat_map5_1 = heatmap(res, vega_5)
heat_map5_2 = heatmap(res, vega_5)
baseline_png5 = png_path + "heat_map_nyc_5.png"
save_png(baseline5, baseline_png5)
save_png(heat_map5_1, png_path + "test_heat_map_nyc_5-1.png")
save_png(heat_map5_2, png_path + "test_heat_map_nyc_5-2.png")
spark.catalog.dropGlobalTempView("nyc_taxi")
assert run_diff_png(baseline_png1, png_path + "test_heat_map_nyc_1-1.png", 0.1)
assert run_diff_png(baseline_png1, png_path + "test_heat_map_nyc_1-2.png", 0.1)
assert run_diff_png(baseline_png2, png_path + "test_heat_map_nyc_2-1.png", 0.1)
assert run_diff_png(baseline_png2, png_path + "test_heat_map_nyc_2-2.png", 0.1)
assert run_diff_png(baseline_png3, png_path + "test_heat_map_nyc_3-1.png", 0.15)
assert run_diff_png(baseline_png3, png_path + "test_heat_map_nyc_3-2.png", 0.15)
assert run_diff_png(baseline_png4, png_path + "test_heat_map_nyc_4-1.png", 0.1)
assert run_diff_png(baseline_png4, png_path + "test_heat_map_nyc_4-2.png", 0.1)
assert run_diff_png(baseline_png5, png_path + "test_heat_map_nyc_5-1.png", 0.2)
assert run_diff_png(baseline_png5, png_path + "test_heat_map_nyc_5-2.png", 0.2)
