fastbook 09 (DT, RF & DL) with California Housing Prices

from fastai.imports import *
from fastai.tabular.all import *
from kaggle import api

data = 'camnugent/california-housing-prices'
api.dataset_list_cli(search=data)

ref                                  title                       size  lastUpdated          downloadCount  voteCount  usabilityRating  
-----------------------------------  -------------------------  -----  -------------------  -------------  ---------  ---------------  
camnugent/california-housing-prices  California Housing Prices  400KB  2017-11-24 03:14:59          70928        741  0.85294116       

path = Path(data.split('/')[1])
api.dataset_download_files(data, path)
path.ls()

(#2) [Path('california-housing-prices/california-housing-prices.zip'),Path('california-housing-prices/housing.csv')]

import zipfile
zipfile.ZipFile(path.ls()[0]).extractall(path)
path.ls()

(#2) [Path('california-housing-prices/california-housing-prices.zip'),Path('california-housing-prices/housing.csv')]

df = pd.read_csv(path/'housing.csv', low_memory=False)
df.head()

	longitude	latitude	housing_median_age	total_rooms	total_bedrooms	population	households	median_income	median_house_value	ocean_proximity
0	-122.23	37.88	41.0	880.0	129.0	322.0	126.0	8.3252	452600.0	NEAR BAY
1	-122.22	37.86	21.0	7099.0	1106.0	2401.0	1138.0	8.3014	358500.0	NEAR BAY
2	-122.24	37.85	52.0	1467.0	190.0	496.0	177.0	7.2574	352100.0	NEAR BAY
3	-122.25	37.85	52.0	1274.0	235.0	558.0	219.0	5.6431	341300.0	NEAR BAY
4	-122.25	37.85	52.0	1627.0	280.0	565.0	259.0	3.8462	342200.0	NEAR BAY

df.columns

Index(['longitude', 'latitude', 'housing_median_age', 'total_rooms',
       'total_bedrooms', 'population', 'households', 'median_income',
       'median_house_value', 'ocean_proximity'],
      dtype='object')

df.hist(figsize=(12,12))

array([[<AxesSubplot:title={'center':'longitude'}>,
        <AxesSubplot:title={'center':'latitude'}>,
        <AxesSubplot:title={'center':'housing_median_age'}>],
       [<AxesSubplot:title={'center':'total_rooms'}>,
        <AxesSubplot:title={'center':'total_bedrooms'}>,
        <AxesSubplot:title={'center':'population'}>],
       [<AxesSubplot:title={'center':'households'}>,
        <AxesSubplot:title={'center':'median_income'}>,
        <AxesSubplot:title={'center':'median_house_value'}>]],
      dtype=object)

x = "total_rooms total_bedrooms population households".split()
df[x] = np.log(df[x])
df.hist(figsize=(12,12))

array([[<AxesSubplot:title={'center':'longitude'}>,
        <AxesSubplot:title={'center':'latitude'}>,
        <AxesSubplot:title={'center':'housing_median_age'}>],
       [<AxesSubplot:title={'center':'total_rooms'}>,
        <AxesSubplot:title={'center':'total_bedrooms'}>,
        <AxesSubplot:title={'center':'population'}>],
       [<AxesSubplot:title={'center':'households'}>,
        <AxesSubplot:title={'center':'median_income'}>,
        <AxesSubplot:title={'center':'median_house_value'}>]],
      dtype=object)

cat = ['ocean_proximity']
cont = ['longitude', 'latitude', 'housing_median_age', 'total_rooms', 'total_bedrooms', 'population', 'households', 'median_income']
procs = [Categorify, FillMissing, Normalize]
to = TabularPandas(df, procs, cat, cont, 'median_house_value', RegressionBlock(),  RandomSplitter()(df), reduce_memory=False).dataloaders(path='.')

xs,y = to.train.xs,to.train.y
val_xs,val_y = to.valid.xs,to.valid.y

RF

from sklearn.ensemble import RandomForestRegressor

m = RandomForestRegressor(100, min_samples_leaf=5).fit(xs, y)
print('MAE:', abs(val_y - m.predict(val_xs)).mean())

x = pd.DataFrame({'cols':xs.columns, 'imp':m.feature_importances_}).sort_values('imp', ascending=False)
x.set_index('cols').plot(kind='barh')
x

MAE: 32150.47857078855

	cols	imp
9	median_income	0.551278
0	ocean_proximity	0.117905
2	longitude	0.108562
3	latitude	0.106565
4	housing_median_age	0.046819
7	population	0.023806
5	total_rooms	0.017090
6	total_bedrooms	0.014583
8	households	0.013367
1	total_bedrooms_na	0.000025

DT

from sklearn.tree import DecisionTreeRegressor, export_graphviz

m = DecisionTreeRegressor(max_leaf_nodes=30).fit(xs, y)
preds = m.predict(val_xs)
abs(val_y - preds).mean()

48784.69675117493

import graphviz

def draw_tree(t, df, size=10, ratio=0.6, precision=2, **kwargs):
    s=export_graphviz(t, out_file=None, feature_names=df.columns, filled=True, rounded=True,
                      special_characters=True, rotate=False, precision=precision, **kwargs)
    return graphviz.Source(re.sub('Tree {', f'Tree {{ size={size}; ratio={ratio}', s))

draw_tree(m, xs, size=10)

<!DOCTYPE svg PUBLIC "-//W3C//DTD SVG 1.1//EN" "http://www.w3.org/Graphics/SVG/1.1/DTD/svg11.dtd">

DNN

learn = tabular_learner(to, metrics=L1LossFlat(), layers=[10,10])
learn.lr_find(suggest_funcs=(slide, valley))

SuggestedLRs(slide=6.309573450380412e-07, valley=0.17378008365631104)

learn.fit(10, lr=0.1)
learn.recorder.plot_loss()

epoch	train_loss	valid_loss	None	time
0	52540518400.000000	49501884416.000000	195610.734375	00:02
1	41421684736.000000	36557541376.000000	169839.937500	00:02
2	30079684608.000000	28439451648.000000	151214.703125	00:02
3	20997189632.000000	15527985152.000000	108336.179688	00:02
4	14592946176.000000	18044657664.000000	120398.703125	00:02
5	10348133376.000000	7776253952.000000	70381.062500	00:02
6	7479977984.000000	6708659200.000000	64639.753906	00:02
7	5428857344.000000	4700456960.000000	49034.535156	00:02
8	4591148032.000000	3880753408.000000	42761.687500	00:02
9	4247306496.000000	3972862464.000000	43510.292969	00:02