In [1]:
import numpy as np
import sklearn
In [2]:
names = ["Mike","Max","Jörg","Simon","Jana","Felix","Markus"]
In [3]:
np.random.shuffle(names)
print(names)

['Jana', 'Markus', 'Jörg', 'Mike', 'Simon', 'Felix', 'Max']
In [4]:
import pandas as pd
In [5]:
data = pd.read_csv("../14_trees/train.tsv", sep = "\t").dropna()
In [6]:
data
Out[6]:
species latitude longitude stem_diameter_cm height_m crown_radius_m
0 Fagus sylvatica 46.01 11.44 5.0 6.7 2.05
1 Fagus sylvatica 46.03 11.44 5.0 11.1 2.90
2 Fagus sylvatica 46.03 11.44 5.0 14.2 4.00
3 Fagus sylvatica 46.03 11.44 5.0 5.7 2.10
4 Fagus sylvatica 46.03 11.44 5.0 8.0 4.15
... ... ... ... ... ... ...
35094 Fagus sylvatica 47.38 9.09 36.0 22.6 4.50
35095 Fagus sylvatica 46.74 6.85 37.0 34.3 4.50
35096 Fagus sylvatica 47.45 9.29 38.0 33.8 5.00
35097 Fagus sylvatica 46.52 6.35 40.0 29.3 6.00
35098 Fagus sylvatica 47.31 8.99 59.0 36.9 7.00

31311 rows × 6 columns

In [7]:
y = pd.read_csv("../14_trees/train.tsv", sep = "\t")
In [8]:
y = data["species"]
In [9]:
print(y)

0        Fagus sylvatica
1        Fagus sylvatica
2        Fagus sylvatica
3        Fagus sylvatica
4        Fagus sylvatica
              ...       
35094    Fagus sylvatica
35095    Fagus sylvatica
35096    Fagus sylvatica
35097    Fagus sylvatica
35098    Fagus sylvatica
Name: species, Length: 31311, dtype: object
In [10]:
X = data[["latitude", "longitude", "stem_diameter_cm", "height_m", "crown_radius_m"]]
X
Out[10]:
latitude longitude stem_diameter_cm height_m crown_radius_m
0 46.01 11.44 5.0 6.7 2.05
1 46.03 11.44 5.0 11.1 2.90
2 46.03 11.44 5.0 14.2 4.00
3 46.03 11.44 5.0 5.7 2.10
4 46.03 11.44 5.0 8.0 4.15
... ... ... ... ... ...
35094 47.38 9.09 36.0 22.6 4.50
35095 46.74 6.85 37.0 34.3 4.50
35096 47.45 9.29 38.0 33.8 5.00
35097 46.52 6.35 40.0 29.3 6.00
35098 47.31 8.99 59.0 36.9 7.00

31311 rows × 5 columns

In [11]:
X = X.to_numpy()
In [12]:
y = y.to_numpy()
In [13]:
np.unique(y, return_counts = True)
Out[13]:
(array(['Fagus sylvatica', 'Pinus pinaster', 'Quercus ilex'], dtype=object),
 array([ 5194,  9417, 16700]))
In [14]:
test = pd.read_csv("../14_trees/test.tsv", sep = "\t")
In [15]:
def majority_prediction(X, y, test):
    u = np.unique(y, return_counts = True)
    majority = u[0][u[1].argmax()]
    prediction = pd.DataFrame()
    prediction["tree_id"] = test.tree_id
    prediction["species"] = majority
    return prediction
In [16]:
majority_prediction(X, y, test).to_csv("tree_pred.tsv", sep='\t', index=False)
In [17]:
from sklearn.neighbors import KNeighborsClassifier
neigh = KNeighborsClassifier(n_neighbors=8)
neigh.fit(X, y)
Out[17]:
KNeighborsClassifier(n_neighbors=8)
In a Jupyter environment, please rerun this cell to show the HTML representation or trust the notebook.
On GitHub, the HTML representation is unable to render, please try loading this page with nbviewer.org.
KNeighborsClassifier(n_neighbors=8)
In [18]:
print(neigh.predict(X))

['Quercus ilex' 'Fagus sylvatica' 'Fagus sylvatica' ... 'Fagus sylvatica'
 'Fagus sylvatica' 'Fagus sylvatica']
In [19]:
neigh.score(X, y)
Out[19]:
0.9363482482194756
In [20]:
test
Out[20]:
tree_id latitude longitude stem_diameter_cm height_m crown_radius_m
0 T_75102 46.01 11.44 5.0 14.2 3.10
1 T_75103 46.01 11.44 5.0 5.4 1.80
2 T_75118 46.03 11.43 5.0 4.9 1.75
3 T_75152 46.05 10.99 5.0 6.5 1.55
4 T_75161 46.05 10.99 5.0 4.9 1.90
... ... ... ... ... ... ...
8892 T_497140 46.24 7.26 21.0 2.6 3.00
8893 T_497324 47.39 7.36 32.0 21.6 4.00
8894 T_497361 46.09 8.99 36.0 3.0 3.50
8895 T_497401 46.74 6.96 37.0 32.3 4.50
8896 T_497446 47.32 7.26 38.0 28.2 4.00

8897 rows × 6 columns

In [21]:
test_pd = test[["latitude", "longitude", "stem_diameter_cm", "height_m", "crown_radius_m"]].dropna()
In [22]:
prediction = pd.DataFrame()
prediction["tree_id"] = test.dropna().tree_id
prediction["species"] = neigh.predict(test_pd)

/home/markus/miniconda3/envs/jupyter/lib/python3.9/site-packages/sklearn/base.py:443: UserWarning: X has feature names, but KNeighborsClassifier was fitted without feature names
  warnings.warn(
In [23]:
prediction.to_csv("species_pred_knn8.tsv",sep='\t', index=False)
In [24]:
from sklearn.model_selection import train_test_split
In [25]:
X_train, X_val, y_train, y_val = train_test_split(X, y)
In [26]:
X_train.shape, X_val.shape
Out[26]:
((23483, 5), (7828, 5))
In [27]:
scores = []
for k in range(1, 20):
    neigh = KNeighborsClassifier(n_neighbors=k)
    neigh.fit(X_train, y_train)
    scores.append(neigh.score(X_val, y_val))
    
import matplotlib.pyplot as plt
plt.plot(list(range(1,len(scores)+1)), scores)
Out[27]:
[<matplotlib.lines.Line2D at 0x7f183111a730>]