Master Pandas: From Installation to Advanced Data Manipulation and Stock Screening

Installation

Pandas can be installed easily via Anaconda (pre‑installed) or with pip: pip install pandas Import the library before use:

import pandas as pd

Data Structures

Pandas provides three main data structures built on NumPy:

Series : a one‑dimensional array with an index.

DataFrame : a two‑dimensional table with rows and columns.

Panel : a three‑dimensional container (rarely used).

Series

Example of creating a Series with random numbers:

import pandas as pd
import numpy as np
s = pd.Series(np.random.randn(4))
print(s)

The default index is shown in the output table.

DataFrame

Create a DataFrame from a dictionary of equal‑length lists:

data = {
    'name': ['张三', '李四', '王五'],
    'gender': ['M', 'F', 'M'],
    'height': [174, 160, 185],
    'weight': [80, 48, 70]
}
frame = pd.DataFrame(data)
print(frame)

DataFrames support column ordering via the columns keyword and custom indexing via index:

frame2 = pd.DataFrame(data, columns=['name', 'gender', 'weight'], index=['one', 'two', 'three'])
print(frame2)

Data Access and Traversal

Access rows by integer location:

frame2.iloc[0]
# name      张三
gender     M
weight    80

Access rows by label:

frame2.loc['two']
# name      李四
gender     F
weight    48

Iterate over rows:

for i in range(len(frame2)):
    print(frame2.iloc[i])

for index, row in frame2.iterrows():
    print(row)

Adding and Deleting Columns

Add a new column (e.g., BMI) calculated from existing columns:

frame['BMI'] = frame['weight'] / (frame['height'] * frame['height']) * 10000
print(frame)

Delete a column:

del frame2['gender']
print(frame2)

Adding and Deleting Rows

Add rows by creating a new DataFrame and appending:

frame3 = pd.DataFrame([
    ['小红', 46],
    ['小明', 68]
], columns=['name', 'weight'], index=['four', 'five'])
frame4 = frame2.append(frame3)
print(frame4)

Delete rows by index label:

frame4 = frame4.drop('four')
print(frame4)

Data Filtering

Select the first two records:

frame[:2]

Filter rows where a condition holds, e.g., BMI > 20:

mask = frame['BMI'] > 20
filtered = frame.loc[mask]
print(filtered)

Panel (Brief)

Panel is a three‑dimensional structure similar to a dictionary of DataFrames; it is rarely used and omitted here.

Practical Stock Screening with Pandas

Using Pandas together with TuShare and TALib to select A‑share stocks that meet specific performance criteria:

import tushare as ts
import talib as tl

data = ts.get_k_data('300573', autype='qfq')
# Calculate daily rate of change
data['p_change'] = tl.ROC(data['close'], 1)

threshold = 60
if len(data) < threshold:
    return False

data = data.tail(n=threshold)
ratio_increase = (data.iloc[-1]['close'] - data.iloc[0]['close']) / data.iloc[0]['close']
if ratio_increase < 0.6:
    return False

for i in range(1, len(data)):
    # Single‑day drop >7%
    if data.iloc[i]['p_change'] < -7:
        return False
    # Two‑day cumulative drop >10%
    if (data.iloc[i]['p_change'] + data.iloc[i-1]['p_change']) < -10:
        return False
return True

The script returns a list of qualifying stocks, demonstrating how Pandas can power quantitative finance workflows.