Scribd
Upload
3 views11 pages

LogisticRegression RST

Uploaded by

Fahad Nasim
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as TXT, PDF, TXT or read online on Scribd
3 views11 pages

LogisticRegression RST

Uploaded by

Fahad Nasim
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as TXT, PDF, TXT or read online on Scribd
You are on page 1/ 11

..

code:: ipython3

import numpy as np
import pandas as pd
import plotly.express as px

.. code:: ipython3

df = pd.read_csv(r"C:\Users\DELL\Downloads\archive\loan_data.csv")

.. code:: ipython3

ds = df.copy()
ds.drop(columns=['person_home_ownership', 'loan_intent'], inplace = True)

.. code:: ipython3

ds = ds.sample(frac = 1, random_state = 42).reset_index(drop = True)


ds

.. raw:: html

<div>
<style scoped>
.dataframe tbody tr th:only-of-type {
vertical-align: middle;
}

.dataframe tbody tr th {
vertical-align: top;
}

.dataframe thead th {
text-align: right;
}
</style>
<table border="1" class="dataframe">
<thead>
<tr style="text-align: right;">
<th></th>
<th>person_age</th>
<th>person_gender</th>
<th>person_education</th>
<th>person_income</th>
<th>person_emp_exp</th>
<th>loan_amnt</th>
<th>loan_int_rate</th>
<th>loan_percent_income</th>
<th>cb_person_cred_hist_length</th>
<th>credit_score</th>
<th>previous_loan_defaults_on_file</th>
<th>loan_status</th>
</tr>
</thead>
<tbody>
<tr>
<th>0</th>
<td>32.0</td>
<td>male</td>
<td>Associate</td>
<td>96865.0</td>
<td>10</td>
<td>7500.0</td>
<td>6.04</td>
<td>0.08</td>
<td>10.0</td>
<td>601</td>
<td>No</td>
<td>0</td>
</tr>
<tr>
<th>1</th>
<td>24.0</td>
<td>male</td>
<td>Associate</td>
<td>56838.0</td>
<td>6</td>
<td>9000.0</td>
<td>11.49</td>
<td>0.16</td>
<td>4.0</td>
<td>647</td>
<td>Yes</td>
<td>0</td>
</tr>
<tr>
<th>2</th>
<td>22.0</td>
<td>female</td>
<td>Master</td>
<td>37298.0</td>
<td>0</td>
<td>5000.0</td>
<td>14.88</td>
<td>0.13</td>
<td>4.0</td>
<td>711</td>
<td>No</td>
<td>1</td>
</tr>
<tr>
<th>3</th>
<td>23.0</td>
<td>female</td>
<td>Bachelor</td>
<td>39944.0</td>
<td>1</td>
<td>5000.0</td>
<td>13.99</td>
<td>0.13</td>
<td>3.0</td>
<td>597</td>
<td>Yes</td>
<td>0</td>
</tr>
<tr>
<th>4</th>
<td>42.0</td>
<td>male</td>
<td>High School</td>
<td>67974.0</td>
<td>20</td>
<td>10000.0</td>
<td>15.41</td>
<td>0.15</td>
<td>11.0</td>
<td>586</td>
<td>No</td>
<td>1</td>
</tr>
<tr>
<th>...</th>
<td>...</td>
<td>...</td>
<td>...</td>
<td>...</td>
<td>...</td>
<td>...</td>
<td>...</td>
<td>...</td>
<td>...</td>
<td>...</td>
<td>...</td>
<td>...</td>
</tr>
<tr>
<th>44995</th>
<td>26.0</td>
<td>male</td>
<td>High School</td>
<td>88451.0</td>
<td>6</td>
<td>10625.0</td>
<td>6.03</td>
<td>0.12</td>
<td>4.0</td>
<td>559</td>
<td>Yes</td>
<td>0</td>
</tr>
<tr>
<th>44996</th>
<td>25.0</td>
<td>female</td>
<td>High School</td>
<td>34772.0</td>
<td>3</td>
<td>5876.0</td>
<td>11.69</td>
<td>0.17</td>
<td>4.0</td>
<td>647</td>
<td>No</td>
<td>1</td>
</tr>
<tr>
<th>44997</th>
<td>33.0</td>
<td>female</td>
<td>Associate</td>
<td>58317.0</td>
<td>9</td>
<td>10707.0</td>
<td>10.74</td>
<td>0.18</td>
<td>9.0</td>
<td>652</td>
<td>Yes</td>
<td>0</td>
</tr>
<tr>
<th>44998</th>
<td>26.0</td>
<td>male</td>
<td>Master</td>
<td>178602.0</td>
<td>6</td>
<td>20000.0</td>
<td>17.99</td>
<td>0.11</td>
<td>3.0</td>
<td>604</td>
<td>No</td>
<td>1</td>
</tr>
<tr>
<th>44999</th>
<td>26.0</td>
<td>female</td>
<td>High School</td>
<td>210894.0</td>
<td>4</td>
<td>15000.0</td>
<td>11.01</td>
<td>0.07</td>
<td>3.0</td>
<td>664</td>
<td>Yes</td>
<td>0</td>
</tr>
</tbody>
</table>
<p>45000 rows × 12 columns</p>
</div>

.. code:: ipython3

ds['person_gender'] = pd.factorize(ds['person_gender'])[0]
ds['previous_loan_defaults_on_file'] =
pd.factorize(ds['previous_loan_defaults_on_file'])[0]
education_hierarchy = {'High School': 0, 'Associate': 1, 'Bachelor':
2,'Master': 3, 'Doctorate': 4}
ds['person_education'] = ds['person_education'].map(education_hierarchy)

.. code:: ipython3

X = ds.iloc[:, [i for i in range(0, 12)]].values


Y = ds.iloc[:, -1].values

.. code:: ipython3

ds

.. raw:: html

<div>
<style scoped>
.dataframe tbody tr th:only-of-type {
vertical-align: middle;
}

.dataframe tbody tr th {
vertical-align: top;
}

.dataframe thead th {
text-align: right;
}
</style>
<table border="1" class="dataframe">
<thead>
<tr style="text-align: right;">
<th></th>
<th>person_age</th>
<th>person_gender</th>
<th>person_education</th>
<th>person_income</th>
<th>person_emp_exp</th>
<th>loan_amnt</th>
<th>loan_int_rate</th>
<th>loan_percent_income</th>
<th>cb_person_cred_hist_length</th>
<th>credit_score</th>
<th>previous_loan_defaults_on_file</th>
<th>loan_status</th>
</tr>
</thead>
<tbody>
<tr>
<th>0</th>
<td>32.0</td>
<td>0</td>
<td>1</td>
<td>96865.0</td>
<td>10</td>
<td>7500.0</td>
<td>6.04</td>
<td>0.08</td>
<td>10.0</td>
<td>601</td>
<td>0</td>
<td>0</td>
</tr>
<tr>
<th>1</th>
<td>24.0</td>
<td>0</td>
<td>1</td>
<td>56838.0</td>
<td>6</td>
<td>9000.0</td>
<td>11.49</td>
<td>0.16</td>
<td>4.0</td>
<td>647</td>
<td>1</td>
<td>0</td>
</tr>
<tr>
<th>2</th>
<td>22.0</td>
<td>1</td>
<td>3</td>
<td>37298.0</td>
<td>0</td>
<td>5000.0</td>
<td>14.88</td>
<td>0.13</td>
<td>4.0</td>
<td>711</td>
<td>0</td>
<td>1</td>
</tr>
<tr>
<th>3</th>
<td>23.0</td>
<td>1</td>
<td>2</td>
<td>39944.0</td>
<td>1</td>
<td>5000.0</td>
<td>13.99</td>
<td>0.13</td>
<td>3.0</td>
<td>597</td>
<td>1</td>
<td>0</td>
</tr>
<tr>
<th>4</th>
<td>42.0</td>
<td>0</td>
<td>0</td>
<td>67974.0</td>
<td>20</td>
<td>10000.0</td>
<td>15.41</td>
<td>0.15</td>
<td>11.0</td>
<td>586</td>
<td>0</td>
<td>1</td>
</tr>
<tr>
<th>...</th>
<td>...</td>
<td>...</td>
<td>...</td>
<td>...</td>
<td>...</td>
<td>...</td>
<td>...</td>
<td>...</td>
<td>...</td>
<td>...</td>
<td>...</td>
<td>...</td>
</tr>
<tr>
<th>44995</th>
<td>26.0</td>
<td>0</td>
<td>0</td>
<td>88451.0</td>
<td>6</td>
<td>10625.0</td>
<td>6.03</td>
<td>0.12</td>
<td>4.0</td>
<td>559</td>
<td>1</td>
<td>0</td>
</tr>
<tr>
<th>44996</th>
<td>25.0</td>
<td>1</td>
<td>0</td>
<td>34772.0</td>
<td>3</td>
<td>5876.0</td>
<td>11.69</td>
<td>0.17</td>
<td>4.0</td>
<td>647</td>
<td>0</td>
<td>1</td>
</tr>
<tr>
<th>44997</th>
<td>33.0</td>
<td>1</td>
<td>1</td>
<td>58317.0</td>
<td>9</td>
<td>10707.0</td>
<td>10.74</td>
<td>0.18</td>
<td>9.0</td>
<td>652</td>
<td>1</td>
<td>0</td>
</tr>
<tr>
<th>44998</th>
<td>26.0</td>
<td>0</td>
<td>3</td>
<td>178602.0</td>
<td>6</td>
<td>20000.0</td>
<td>17.99</td>
<td>0.11</td>
<td>3.0</td>
<td>604</td>
<td>0</td>
<td>1</td>
</tr>
<tr>
<th>44999</th>
<td>26.0</td>
<td>1</td>
<td>0</td>
<td>210894.0</td>
<td>4</td>
<td>15000.0</td>
<td>11.01</td>
<td>0.07</td>
<td>3.0</td>
<td>664</td>
<td>1</td>
<td>0</td>
</tr>
</tbody>
</table>
<p>45000 rows × 12 columns</p>
</div>

.. code:: ipython3

ds.dtypes

.. parsed-literal::

person_age float64
person_gender int64
person_education int64
person_income float64
person_emp_exp int64
loan_amnt float64
loan_int_rate float64
loan_percent_income float64
cb_person_cred_hist_length float64
credit_score int64
previous_loan_defaults_on_file int64
loan_status int64
dtype: object

.. code:: ipython3

def feature_scaling(X):
for i in range(12):
if X[:, i].max() == 1:
continue
else:
X[:, i] = (X[:, i] - X[:, i].min())/(X[:, i].max() - X[:, i].min())
return X
'''def feature_scaling(X):
X = X.astype(float) # Ensure float for scaling

col_max = X.max(axis=0)
col_min = X.min(axis=0)

# Boolean mask: True for binary columns (only 0 and 1)


binary_cols = (col_min == 0) & (col_max == 1)

# Avoid division by zero


range_vals = col_max - col_min
range_vals[range_vals == 0] = 1

# Perform min-max scaling


X_scaled = (X - col_min) / range_vals

# Put back the binary columns without scaling


X_scaled[:, binary_cols] = X[:, binary_cols]

return X_scaled
'''

.. parsed-literal::

'def feature_scaling(X):\n X = X.astype(float) # Ensure float for scaling\


n\n col_max = X.max(axis=0)\n col_min = X.min(axis=0)\n\n # Boolean mask:
True for binary columns (only 0 and 1)\n binary_cols = (col_min == 0) & (col_max
== 1)\n\n # Avoid division by zero\n range_vals = col_max - col_min\n
range_vals[range_vals == 0] = 1\n\n # Perform min-max scaling\n X_scaled = (X
- col_min) / range_vals\n\n # Put back the binary columns without scaling\n
X_scaled[:, binary_cols] = X[:, binary_cols]\n\n return X_scaled\n'

.. code:: ipython3
X1 = X.copy()

.. code:: ipython3

X1 = feature_scaling(X1)

.. code:: ipython3

training = X1[:3600]
x = training[:, :-1]
y = training[:, -1].reshape(3600, 1)
test = X1[3601:]

.. code:: ipython3

def sigmoid(f):
return 1/(1+(np.exp(-f)))

.. code:: ipython3

w = np.ones(11).reshape(11, 1); b = 0; m = len(x); f = np.dot(x, w) + b


s = sigmoid(f)

.. code:: ipython3

def cost_calc(y, s):


loss = -(y*np.log(s) + (1 - s)*np.log(1- s))
cost = (1/m)*np.sum(loss)
return cost

.. code:: ipython3

def derivative_calc(x, y, s):


dj_dw = (1/m)*np.dot(x.T, (s - y))
dj_db = (1/m)*np.sum(s - y)
return ([dj_dw, dj_db])

.. code:: ipython3

derivative_calc(x, y, s)

.. parsed-literal::

[array([[0.0452881 ],
[0.31779622],
[0.23901949],
[0.00751574],
[0.0321477 ],
[0.18264943],
[0.23971088],
[0.13128926],
[0.10121809],
[0.37012421],
[0.47815916]]),
np.float64(0.6954458548817282)]
.. code:: ipython3

def gradient_descent(w, b, a):


d = derivative_calc(x, y, s)
for i in range(10000):
w = w - a*d[0]
b = b - a*d[1]
return (w, b)

.. code:: ipython3

gradient_descent(w, b, 0.001)

.. parsed-literal::

(array([[ 0.54711902],
[-2.17796218],
[-1.3901949 ],
[ 0.9248426 ],
[ 0.67852302],
[-0.8264943 ],
[-1.39710879],
[-0.31289259],
[-0.01218094],
[-2.70124215],
[-3.78159158]]),
np.float64(-6.954458548815924))

You might also like