Have you ever wondered how data science can unlock the potential for economic growth in a country like Nepal? As someone who has worked with economic data, I've seen firsthand the challenges of making sense of complex national data, which often presents itself in silos, making it difficult to extract actionable intelligence for informed decision-making. However, with a structured data science approach, we can move beyond mere observation to building predictive models that offer genuine foresight. In this guide, I'll walk you through how to access, clean, engineer features from, and finally build a simple machine learning model using publicly available World Bank data to forecast key economic indicators for Nepal. You'll learn not just the technical steps, but also the critical thought process behind transforming raw numbers into insights that can truly drive economic growth and development.
Key Takeaways
- Robust API integration is crucial for accessing reliable economic data.
- Data cleaning and feature engineering are essential steps in preparing data for machine learning models.
- A simple machine learning model can be used to forecast key economic indicators, such as GDP per capita and inflation rate.
The Problem
Nepal's economy is rapidly growing, but there is a need for data-driven insights to inform policy decisions and drive development. Existing economic data is often siloed and difficult to analyze, hindering the ability of policymakers and developers to make informed decisions.
Data and Sources
The World Bank's Open Data API provides publicly available economic data, including GDP per capita and inflation rate data for Nepal. The data can be accessed through the API endpoint: https://api.worldbank.org/v2/country/NPL/indicator/NY.GDP.PCAP.CD?format=json&date=1960:2022. A freshness note: Data accessed on 2026-07-25.
Step 1 — Data Retrieval
The first step is to fetch the necessary data from the World Bank's API. We can use the `requests` library in Python to send a GET request to the API endpoint.
import requests
response = requests.get("https://api.worldbank.org/v2/country/NPL/indicator/NY.GDP.PCAP.CD?format=json&date=1960:2022")
data = response.json()
Step 2 — Data Cleaning
After fetching the data, we need to clean it by handling missing values and outliers. We can use the `pandas` library to create a DataFrame and clean the data.
import pandas as pd
df = pd.DataFrame(data[1])
df = df.dropna() # drop rows with missing values
Step 3 — Feature Engineering
Next, we need to engineer features from the existing data. We can create new features, such as the average GDP per capita over a certain period.
df['avg_gdp'] = df['value'].rolling(window=5).mean()
Step 4 — Model Selection and Training
Finally, we can select a suitable machine learning model and train it on the data. We can use a simple linear regression model to forecast the GDP per capita.
from sklearn.linear_model import LinearRegression
X = df[['avg_gdp']]
y = df['value']
model = LinearRegression()
model.fit(X, y)
Complete Script
The full runnable script combining all steps:
#!/usr/bin/env python3
import requests
import pandas as pd
from sklearn.linear_model import LinearRegression
def fetch_data():
response = requests.get("https://api.worldbank.org/v2/country/NPL/indicator/NY.GDP.PCAP.CD?format=json&date=1960:2022")
data = response.json()
return data
def clean_data(data):
df = pd.DataFrame(data[1])
df = df.dropna() # drop rows with missing values
return df
def engineer_features(df):
df['avg_gdp'] = df['value'].rolling(window=5).mean()
return df
def train_model(df):
X = df[['avg_gdp']]
y = df['value']
model = LinearRegression()
model.fit(X, y)
return model
if __name__ == "__main__":
data = fetch_data()
df = clean_data(data)
df = engineer_features(df)
model = train_model(df)
print(model.predict([[1000]])) # forecast GDP per capita
Expected Output
The script will output the forecasted GDP per capita for Nepal.
Limitations and Tradeoffs
This approach has several limitations, including the assumption that the relationship between the features and the target variable is linear. Additionally, the model is sensitive to outliers and missing values in the data. For a more robust model, we could use a more advanced machine learning algorithm, such as a decision tree or random forest, and handle outliers and missing values more effectively.
Frequently Asked Questions
What is the purpose of feature engineering in this script?
Feature engineering is used to create new features from the existing data, such as the average GDP per capita over a certain period. This helps to improve the accuracy of the model by providing more relevant and informative features.
How can I handle outliers and missing values in the data?
Outliers and missing values can be handled using various techniques, such as dropping rows with missing values, imputing missing values with the mean or median, or using a more robust machine learning algorithm that can handle outliers and missing values.
Can I use a more advanced machine learning algorithm for this problem?
Yes, a more advanced machine learning algorithm, such as a decision tree or random forest, can be used to improve the accuracy of the model. However, this would require more data and computational resources.
What I'd Change
In conclusion, while this script provides a basic framework for forecasting GDP per capita using machine learning, there are several areas for improvement. For a more robust model, I would use a more advanced machine learning algorithm, handle outliers and missing values more effectively, and incorporate more features, such as inflation rate and population growth. Additionally, I would use more recent data and consider using techniques such as cross-validation to evaluate the model's performance.
