In [1]:

import pandas as pd

import numpy as np

import matplotlib.pyplot as plt

import seaborn as sns

from sklearn.model_selection import train_test_split

from sklearn.ensemble import RandomForestRegressor

from prophet import Prophet

In [2]:

# Load the San Francisco housing data from S3

import boto3

s3 = boto3.client('s3')

obj = s3.get_object(

Bucket='housing-data',

Key='sf_housing_prices.csv'

)

df = pd.read_csv(obj['Body'])

In [3]:

# Preview the data

df.head()

In [4]:

# Filter for Potrero Hill neighborhood

potrero_data = df[df['neighborhood'] == 'Potrero Hill']

potrero_data.shape

In [5]:

# Visualize price trends over time

potrero_data['date'] = pd.to_datetime(potrero_data['date'])

potrero_data.set_index('date', inplace=True)

monthly_avg = potrero_data.resample('M').mean()

plt.figure(figsize=(10, 6))

plt.plot(monthly_avg['price'])

plt.title('Potrero Hill Average Monthly Home Prices')

plt.ylabel('Price ($)')

plt.show()

In [6]:

agent = df.Agent(data=potrero_data)

agent.ask("Predict prices for Q3 2025")