Random sampling

Analyzing Survey Data in Python

EbunOluwa Andrew

Data Scientist

Sampling in survey data analysis

Sampling = small set from large population
- Make inferences about larger population
- Sampling -> manageable form
- Sampling -> sampling error
- Minimize error with large population

Photo by Patrick Fore on Unsplash - kid holding candies

Random sampling

Each member has equal chance of being selected
Reduces bias
High internal validity
High external validity

lottery tickets

.sample() method

DataFrame.sample(n = _None_, frac = _None_, random_state = _None_)
n = number of items to sample
frac = proportion (out of 1) of items to return
random_state = seed number to produce reproducible results

Random sampling example

import pandas as pd
survey = pd.read_csv('ABC_survey.csv')

sample = survey.sample(n=100)
print(sample)

|       | employee_id | gender | onsite_work |
|-------|-------------|--------|-------------|
| 3244  | fffe330     | Female | Yes         |
| 21339 | fffe310     | Male   | Yes         |
| 1122  | fffe390     | Male   | Yes         |
| 4363  | fffe313     | Female | Yes         |

Random sampling example

import pandas as pd
survey = pd.read_csv('ABC_survey.csv')

sample = survey.sample(frac = 0.1)
print(sample)

|     | employee_id | gender | onsite_work |
|-----|-------------|--------|-------------|
| 142 | fffe800     | Female | Yes         |
| 710 | fffe900     | Female | Yes         |
| 242 | fffe700     | Female | Yes         |
| 114 | fffe600     | Female | Yes         |

Random sampling example

import pandas as pd
survey = pd.read_csv('ABC_survey.csv')

sample = survey.sample(
  n = 100, random_state = 123)

import pandas as pd
survey = pd.read_csv('ABC_survey.csv')

sample = survey.sample(
  frac = 0.1, random_state = 123)

|       | employee_id | gender | onsite_work |
|-------|-------------|--------|-------------|
| 21383 | fffe3       | Female | Yes         |
| 82    | fffe0       | Male   | Yes         |
| 20739 | fffe2       | Male   | Yes         |
| 7662  | fffe9       | Female | Yes         |

|       | employee_id | gender | onsite_work |
|-------|-------------|--------|-------------|
| 21383 | fffe3       | Female | Yes         |
| 82    | fffe0       | Male   | Yes         |
| 20739 | fffe2       | Male   | Yes         |
| 7662  | fffe9       | Female | Yes         |

Let's practice!

Analyzing Survey Data in Python