Download as pptx, pdf, or txt
Download as pptx, pdf, or txt
You are on page 1of 17

Unveiling the Future :

PREDICTIVE
ANALYTICS

Presented by : Group 5
Abhilash Hegde
Divya D
Deekshitha
Apoorve
Jefrin Jijo
Introduction to
Predictive Analytics

Making future predictions about unknown events with the help of techniques from data
mining, statistics, machine learning, math modeling, and artificial intelligence is known
as predictive analytics.

For example, predicting sales of an item (say flowers) in a market for a particular day. If it is
valentines day, the sales of roses would be high! We can easily say that the sales of flowers
would be higher on festive days than on regular days.

The insights from the data include patterns, the relationship among different factors that
might be previously unknown.
Why Predictive Analytics?
1 Data Collection
Understanding the different sources of data and how to collect relevant information
is crucial for the success of predictive analytics.

2 Real-World Application
By predicting future outcomes, businesses can make more informed decisions,
optimize processes, and gain a competitive edge in the market.

3 Exploratory Data Analysis (EDA)


Exploring and analyzing the data helps to identify patterns, trends, and outliers,
providing valuable insights for predictive modeling.
Applications

Visualization of Data Data Analysis Machine Learning


Techniques
Using visual representations like Performing rigorous analysis on
charts and graphs to explore the dataset to understand its Applying various machine
patterns and relationships within structure, variables, and learning algorithms to extract
the dataset. significance for predictive meaningful insights and build
modeling. predictive models.
Real World Application of
Predictive Analytics
1 Healthcare 2 Finance
Predictive analytics can be used to In finance, predictive analytics is
forecast patient outcomes, enhance employed for risk assessment, fraud
treatment plans, and improve overall detection, and algorithmic trading
healthcare operations. strategies.

3 Marketing
Businesses utilize predictive analytics to optimize marketing campaigns, personalize
customer experiences, and drive sales growth.
Data Collection for Predictive
Analytics
Digital Footprint Sensor Data
Collecting information from IoT devices
Utilizing online and social media data to
and sensors to monitor and analyze various
gather insights on consumer behavior and
aspects of our environment.
preferences.

Customer Interactions
Tracking and analyzing customer interactions across different touchpoints to understand their
needs and behavior.
Steps for Predictive Analytics
Data Preprocessing
Prepare the dataset by cleaning, transforming, and normalizing it for analysis.

Model Development
Develop and train predictive models using algorithms such as regression,
classification, and clustering.

Evaluation & Validation


Assess the model's performance, validate its accuracy, and optimize it for real-world
application.
Example Dataset

Demand (Million Supply (Million


Date Barrels/Day) Barrels/Day) Price (INR/Liter)
2023-10-01 20 18 101.98
2023-11-01 21 19 105.30
2023-12-01 22 20 102.85
2024-01-01 20 19 102.38
2024-02-01 22 20 ?
Code :
Output:
Exploratory Data Analysis (EDA)
Descriptive Data Visualization Outlier Detection
Statistics
Creating visual summaries Identifying and handling
Using measures like mean, such as histograms, box plots, anomalies that may affect the
median, and standard and scatter plots to uncover accuracy of predictive models.
deviation to describe the main underlying patterns.
features of the dataset.
Feature Engineering in Predictive
Analytics
Feature Creation Dimensionality Reduction

Generating new features from existing data to Reducing the number of input variables by selecting
improve the model's predictive performance. the most relevant features for the model.
Model Selection in Predictive
Analytics
1 Regression Models 2 Classification Models
Used for predicting continuous values, Applied to categorize data into distinct
such as sales forecasts or temperature classes, like identifying fraud or disease
predictions. diagnosis.

3 Ensemble Techniques
Utilizing multiple models to make predictions for improved accuracy and robustness.
Challenges & Considerations

2. Overfitting
1. Data Quality - Use regularization techniques or feature selection
- Implement data cleaning techniques such as handling missing methods in machine learning models to prevent overfitting
values and outliers to ensure the quality of input data. and enhance generalization.

# Handling missing values


df = df.dropna() # Drop rows with missing values
# or from sklearn.linear_model import Ridge
df['column_name'].fillna(df['column_name'].mean(), model = Ridge(alpha=1.0) # Adjust the alpha
inplace=True) # Fill missing values with mean parameter for regularization strength
model.fit(X_train, y_train)
# Handling outliers
from scipy.stats import zscore
df['z_score'] = zscore(df['column_name'])
df = df[(df['z_score'] < 3) & (df['z_score'] > -3)] # Keep
data within 3 standard deviations
Challenges & Considerations

3. Model Interpretability:
- Utilize model interpretability tools or techniques like Partial
Dependence Plots (PDP):values for explaining model predictions
to stakeholders. PDPs show the relationship between a feature
and the predicted outcome while keeping other features constant.
They help in understanding the effect of a single variable on the
model's prediction.

from sklearn.inspection import


plot_partial_dependence
plot_partial_dependence(model, X_train, features=[0,
1], grid_resolution=10)
Conclusion
In conclusion, predictive analytics empowers organizations to make informed and proactive
decisions by harnessing the power of data. By utilizing statistical algorithms and machine learning
models, predictive analytics enables the anticipation of future trends, behaviors, and outcomes. This
transformative approach helps businesses optimize processes, enhance resource allocation, and
mitigate risks. Whether applied in finance, marketing, healthcare, or various other domains,
predictive analytics serves as a strategic tool for gaining a competitive edge and fostering
innovation. Embracing this data-driven methodology not only enhances operational efficiency but
also enables organizations to adapt and thrive in an ever-evolving landscape, ultimately leading to
more informed and successful decision-making.

You might also like