EXPERIMENT - 1

Aim:-Study of Power BI Desktop.

Software Used: -Microsoft Power BI.

Theory:-Power BI is a collection of software services, apps, and connectors that work together to turn
your unrelated sources of data into coherent, visually immersive, and interactive insights. Your data
might be an Excel spreadsheet, or a collection of cloud-based and on-premises hybrid data warehouses.
Power BI lets you easily connect to your data sources, visualize and discover what's important, and share
that with anyone or everyone you want.

Procedure:-
Setting up Power BI: -

1. To create a workspace, in the Navigation pane (located at the left),

click Workspaces, and then click Create a Workspace.
2. When prompted to upgrade your account to Power BI Pro, click Try Free.In the Create a Workspace pane (located
at the right), in the Workspace Name box, enter a name for your workspace. The name you enter must be unique
to the tenant. We suggest you name the workspace Sales Analysis, and that you also append your initials. For
example, the name could be Sales Analysis AB.

3. To create the workspace, at the bottom of the pane, click Save.

4. In the Navigation pane, notice that your workspace is open.

5. To open the Power BI Desktop, on the taskbar, click the Microsoft Power BI Desktop shortcut.
6. When prompted, click Sign In.

7. In Power BI Desktop, at the top-right corner, verify that you see your account.
 EXPERIMENT-2

Aim:- Import the data from different sources and load into the target system.

Software Used:-Microsoft Power BI

Theory:-Power BI provides several methods for loading data into the target system, offering flexibility
to cater to diverse data sources and scenarios,
The primary ways include:

1. Direct Query: This method establishes a live connection to the data source, enabling real-time
data analysis without importing it into Power BI. Its is suitable for large datasets where the focus is
on current, up-to-the-minute insights, However,

it relies on the source system for query performance.

2. Import Data: The import option involves loading data directly into the Power BI model,
which resides in-memory. This
approach is effective for smaller to moderately sized datasets, providing faster query performance.
Users can leverage Power

Query Editor for data transformations during the import process.

3. Power Query Editor: A versatile tool within Power BI, Power Query Editor enables users
to connect to various data

sources, apply transformations, and shape the data before loading it into the model. This method
allows for comprehensive

data cleaning. filtering, and structuring.

4. Dataflows: Dataflows facilitate a Power Query experience in the cloud, where data
transformations are applied in a Power

BI service workspace. This provides a reusable and centralized approach for data preparation,
promoting consistency in data
processing across reports and dashboards.

5. Power Automate Integration: Power Bl can integrate with Power Automate (formerly known
as Microsoft Flow) to

automate data loading processes, This integration enables users to trigger data refreshes based on
predefined schedules,
events, or triggers from external systems.

Procedure: -
1. In the Power Query Editor window, in the Queries pane, select the DimEmployee query.

2. To rename the query, in the Query Settings pane (located at the right), in the Name box,
replace the text with Salesperson, and then press Enter. The query name will determine
the model table name. It’s recommended to define concise, yet friendly, names.

3. In the Queries pane, verify that the query name has been updated.

4. To locate a specific column, on the Home ribbon tab, from inside the Manage
Columns group, click the Choose Columns down-arrow, and then select Go
to Column.
5. In the Go to Column window, to order the list by column name, click the AZ sort button,
and then select Name.

6. Select the SalesPersonFlag column, and then click OK.

7. To filter the query, in the SalesPersonFlag column header, click the down-arrow, and
then uncheck FALSE.

8. Click OK
9. In the Query Settings pane, in the Applied Steps list, notice the addition of the Filtered
Rows step.

10. To remove columns, on the Home ribbon tab, from inside the Manage Columns
group, click the Choose Columns icon.

11. . In the Choose Columns window, to uncheck all columns, uncheck the (Select All
Columns) item.

12. . To include columns, check the following six columns:

a. EmployeeKey
b. EmployeeNationalIDAlternateKey
c. FirstName or LastName
d. Title
e. EmailAddress
13. Click OK.
14. . In the Applied Steps list, notice the addition of another query step.

15. To create a single name column, first select the FirstName column header
16. While pressing the Ctrl key, select the LastName column

17. Right-click either of the select column headers, and then in the context menu, select
Merge Columns.

18. In the Merge Columns window, in the Separator dropdown list, select Space.
19. In the New Column Name box, replace the text with Salesperson.
20. Click OK.
21. To rename the
EmployeeNationalIDAlternateKey column,
double-click the
EmployeeNationalIDAlternateKey column
header
22. . Replace the text with EmployeeID, and then press Enter
23. . Use the previous steps to rename the EmailAddress column to UPN.

24. At the bottom-left, in the status bar, verify that the query has five columns and 18 rows
 EXPERIMENT-3

Aim:-Perform the ETL process to construct the database in POWER BI.

Software Used:-Microsoft Power BI.

Theory:-Each loading method in Power Bl aligns with theoretical principles of data integration and
ETL, allowing users to choose the most suitable approach based on their data volume. performance
requirements, and real-time analysis needs. In Power Bl. the

Extract, Transform, Load (ETL) process is a theoretical framework guiding the construction of a
database within the Power BI model.

1. Extract: The process begins with data extraction from various sources, such as
databases, spreadsheets, or cloud services.

Power Bl supports a wide range of connectors, adhering to the principle of extracting diverse data
types. The extraction

aligns with the theoretical foundation of data connectivity, ensuring compatibility with various
source systems.

2. Transform: Power BI's Power Query Editor plays a pivotal role in the transformation phase.
Rooted in theoretical

principles of relational algebra, it allows users to shape and clean the data. The transformations
include filtering, merging,

and aggregating data, ensuring data quality and relevance. The M language, based on functional
programming theory.

facilitates advanced transformations.

3. Load: The transformed data is loaded into the Power Bl model, which follows principles
of relational database theory.

The tabular data model organizes data into tables, leveraging relationships between them. This
optimized structure
adheres to normalization principles, ensuring efficient storage, retrieval, and query performance
within the model.

Procedure: -
1. Open Power BI and Connect to the Data Source
2. Enter the connection settings for the data source. You may be asked for a user ID
and password upon clicking OK.
3. Select the Tables You Need
4. Transform the Data Based on Your Requirements
a. CREATE A NEW QUERY BY REFERENCING AN EXISTING TABLE

b. GROUP THE DATA WITH AN AGGREGATION

i. Click Group By
ii. Then, click Advanced. We need to group using 2 columns.
iii. Click Add grouping. And when a dropdown list appears, select
appropriate column (for example:- product_id)
iv. Then, define the aggregation. Enter Total Premium in the New column
name. Then, select the Operation. And last, select the premium column to
sum.
 v. Finally, click OK to create the grouping.

c. Create model relationships

i. In Power BI Desktop, at the left, click the Model view icon.

ii. If you do not see all seven tables, scroll horizontally to the right, and then drag and
arrange the tables more closely together so they can all be seen at the same time.
iii. To return to the Report view, at the left, click the Report view icon.

iv. To view all table fields, in the Fields pane, right-click an empty area, and then select Expand All.
 v. To create a table visual, in the Fields pane, from inside the Product table, check
the Category field.

vi. To add a column to the table, in the Fields pane, check the Sales | Sales field.
vii. Notice that the table visually lists four product categories, and that the sales value is the
same for each, and the same for the total.

viii. On the Modeling ribbon tab, from inside the Relationships group, click Manage Relationships.

ix. In the Manage Relationships window, notice that no relationships are yet defined.
 x. To create a relationship, click New.
xi. In the Create Relationship window, in the first dropdown list, select the Product table.
xii.Select other appropriate settings like cardinality, filter, etc.
xiii.Click OK.
xiv. In the Manage Relationships window, notice that the new relationship is listed, and
then click Close.
xv. In the report, notice that the table visual has been updated to display different values for
each product category.

xvi. Filters applied to the Product table now propagate to the Sales table
xvii. Switch to Model view, and then notice there is now a connector between the two tables.

xviii. To create a new relationship, from the Reseller table, drag the ResellerKey column on to the
 ResellerKey column of the Sales table.
xix. Create other relationships as required.
xx. Save the Power BI Desktop files.
d. Create Hierarchy
i. In Model view, in the Fields pane, if necessary, expand the Product table.
ii. To create a hierarchy, in the Fields pane, right-click the Category column, and then
select Create Hierarchy.

iii. In the Properties pane (to the left of the Fields pane), in the Name box, replace the
text with Products.
iv. To add the second level to the hierarchy, in the Hierarchy dropdown list, select Subcategory.
v. To add the third level to the hierarchy, in the Hierarchy dropdown list, select Product.
vi. To complete the hierarchy design, click Apply Level Changes.

vii. In the Fields pane, notice the Products hierarchy.

viii. To reveal the hierarchy levels, expand the Products hierarchy.
ix. To organize columns into a display folder, in the Fields pane, first select
the Background Color Format column.
x. While pressing the Ctrl key, select the Font Color Format.
xi. In the Properties pane, in the Display Folder box, enter Formatting.
xii. In the Fields pane, notice that the two columns are now inside a folder
 EXPERIMENT-4

Aim:-Data Visualization from ETL process.

Software Used:-Microsoft Power BI.

Theory:-
Data visualization within the context of Business Intelligence (BI) is intricately tied to the Extract,
Transform, Load (ETL) process, contributing to the effective communication of insights. Theoretical
foundations guide this process:

1. Data Extraction (Extract): In the initial phase of ETL, data is sourced from diverse
systems. Theoretical principles of data compatibility and integration inform decisions on
extracting relevant information from structured and unstructured sources, ensuring data
consistency.
2. Data Transformation (Transform): The transformation phase involves shaping and structuring
the data. Theoretical aspects of data cleaning, normalization, and enrichment are applied,
aligning with principles from relational database theory. Transformation also encompasses the
preparation of data for optimal visual representation.
3. Data Loading (Load): Loaded data is organized within BI systems, often adopting
principles from data modeling theories. A well-structured data model facilitates efficient
querying and supports the creation of meaningful visualizations.
4. Visualization Design: Theoretical concepts from visual perception and cognitive psychology
guide the design of data visualizations. Principles like Edward Tufte's data-ink ratio, color
theory, and Gestalt principles are employed to create visually appealing and informative
dashboards and reports.
5. Interactivity and User Experience: Theoretical underpinnings of human-computer interaction
influence the incorporation of interactive elements. Users can explore data dynamically,
allowing for a more intuitive and engaging experience, aligning with usability and user
experience design theories.

Procedure:-
In power BI Desktop open the Retail analysis sample PBIX file in the report view and select EDIT
There are various methods available in Power BI for visualization, these include :-
There are various methods available in Power BI for visualization, these include :-

Area Chart

1>Select to add a new page.

2>From the Fields pane, select Sales > Last Year Sales, and This Year Sales > Value.
3>Convert the chart to a basic area chart by selecting the Area chart icon from the Visualizations pane.

4>Select Time > FiscalMonth to add it to the Axis well.

5>To display the chart by month, select the ellipses (top right corner of the visual) and
choose Sort by > FiscalMonth. To change the sort order, select the ellipses again and select
either Sort ascending or Sort descending.

Doughnut chart

1>On the Visualizations pane, select the icon for doughnut chart to convert your bar
chart to a doughnut chart. If Last Year Sales isn't in the Values section of the
Visualizations pane, drag it there.
 2>Select Item > Category to add it to the Legend area of the Visualizations pane.

3>Optionally, adjust the size and color of the chart's text.

Cards
1>On the Data pane, expand Store and select the Open Store Count checkbox. By default,
Power BI creates a clustered column chart with a single data value.
You can convert the chart to a card visualization.

2> In the Visualizations pane, select the Card icon.

Combo chart
1>From the Fields pane, select Sales > This Year Sales > Value.
2>Select Sales > Gross Margin This Year and drag it to the Y-axis well.
3>Select Time > Fiscal Month and drag it to the X-axis well.
4>The visualization will be similar to this one.
5>Select the ellipsis again and choose Sort axis > Sort ascending.
6>Convert the column chart to a combo chart. There are two combo charts available:
Line And stacked column and Line and clustered column. With the column chart
selected, from the Visualizations pane select the Line and clustered column chart.

7>In the upper-right corner of the visual, select the More options ellipsis (...) and select
Sort axis >FiscalMonth.

8>From the Fields pane, drag Sales > Last Year Sales to the Line y-axis bucket.
9>Combo chart should look something like this:
 EXPERIMENT-5

Aim:- Implementation of the classification algorithm.

Software Used:-Python

Theory:-
In Business Intelligence (BI), classification algorithms play a crucial role in analyzing and categorizing
data to support data-driven decision-making. These algorithms are rooted in machine learning,
specifically supervised learning, where they learn from labeled historical data to predict the category of
new, unseen data.

1. Training Phase: The classification algorithm is initially trained using a labeled dataset,
learning the associations between input features and specific categories. This process is
grounded in statistical and probabilistic concepts, aiming to identify meaningful patterns and
correlations in the data.
2. Feature Selection and Engineering: Principles of feature selection and engineering are
applied to enhance the model’s capacity to differentiate between classes. Important features are
selected based on their relevance, following information theory and data dimensionality
reduction principles.
3. Model Construction: The algorithm develops a predictive model, often utilizing theoretical
frameworks such as decision trees, support vector machines, or neural networks. These
models embody the learned associations and are used for classifying new data.
4. Evaluation and Validation: Statistical concepts guide the evaluation and validation of
the classification model. Metrics such as precision, recall, and accuracy measure the
model’s effectiveness and ability to generalize to new data.
5. Deployment in BI: Once trained and validated, the classification model is deployed within BI
systems to classify and categorize incoming data. This application of machine learning
enhances BI capabilities, providing insights in areas like customer segmentation, fraud
detection, risk assessment, and other classification-focused business applications.

Procedure:-
import pandas as pd
import matplotlib.pyplot as plt

# Define the rainfall data as a Python list

rainfall = [799, 1174.8, 865.1, 1334.6, 635.4, 918.5, 685.5, 998.6, 784.2, 985, 882.8, 1071]

# Create a pandas DataFrame with a DatetimeIndex to represent the time series data
start_date = "2012-01-01"
end_date = "2013-01-01" # Adjusted to include December 2012
date_range = pd.date_range(start=start_date, end=end_date, freq='M')
rainfall_df = pd.DataFrame(rainfall, index=date_range, columns=["Rainfall"])

# Print the time series data

print(rainfall_df)

# Create a plot of the time series

plt.figure(figsize=(8, 4)) # Define the plot figure size (optional)
plt.plot(rainfall_df.index, rainfall_df['Rainfall'], marker='o', linestyle='-')
plt.xlabel('Date')
plt.ylabel('Rainfall')
plt.title('Monthly Rainfall Time Series')
plt.grid(True)

# Save the plot to a file

plt.savefig("rainfall.png")

# Show the plot

(optional) plt.show()

Output:-
Rainfall
2012-01-31 799.0
2012-02-29 1174.8
2012-03-31 865.1
2012-04-30 1334.6
2012-05-31 635.4
2012-06-30 918.5
2012-07-31 685.5
2012-08-31 998.6
2012-09-30 784.2
2012-10-31 985.0
2012-11-30 882.8
2012-12-31 1071.0
 EXPERIMENT-6

Aim:- Implementation of Decision Tree.

Software Used:- Python

Theory:-A Decision Tree is a supervised learning algorithm useful for both classification and regression
tasks. It visually represents a decision-making process, making it particularly intuitive and easy to
interpret. Key features of Decision Trees are outlined below:

1. Tree Structure: A Decision Tree is made up of nodes, branches, and leaves. Each node signifies
a decision or test on an attribute, each branch indicates an outcome of that decision, and each
leaf represents a class label (in classification) or a predicted value (in regression).
2. Recursive Partitioning: The process of constructing a Decision Tree involves recursively
splitting the dataset into subsets based on the values of the most informative attributes, aiming
to produce subsets that are as homogenous as possible in terms of the target variable.
3. Attribute Selection: At each node, the algorithm chooses the attribute that provides the optimal
split, typically using metrics such as Information Gain (for classification) or Mean Squared
Error (for regression).
4. Pruning: Decision Trees may overfit by capturing noise in the data. Pruning is a technique
used to reduce overfitting by removing branches that do not significantly enhance the tree's
performance on a validation dataset.
5. Interpretability: A major advantage of Decision Trees is their interpretability. The
decision-making process is easily visualized, making them highly valuable for explaining model
predictions.

Procedure:-
import pandas as pd
import matplotlib.pyplot as plt
from sklearn.tree import DecisionTreeClassifier, plot_tree

# Create a sample dataset for the example

data = {
'age': [25, 30, 35, 40, 45, 28, 32, 37, 42, 47],
'shoeSize': [8, 9, 10, 11, 12, 9, 10, 11, 12, 13],
'score': [75, 82, 88, 92, 95, 80, 85, 89, 92, 96],
'nativeSpeaker': ['Yes', 'No', 'Yes', 'No', 'Yes', 'Yes', 'No', 'Yes', 'No', 'Yes']
}

# Create a DataFrame from the sample data

sample_data = pd.DataFrame(data)

# Define the features (predictors) and the target variable

features = ['age', 'shoeSize',
'score'] target = 'nativeSpeaker'

# Split the data into X (features) and y (target variable)

X = sample_data[features]
y = sample_data[target]

# Create a decision tree classifier

decision_tree = DecisionTreeClassifier()
decision_tree.fit(X, y)

# Plot the decision tree

plt.figure(figsize=(10, 6)) # Define the figure size (optional)
plot_tree(decision_tree, filled=True, feature_names=features, class_names=list(y.unique()))
plt.title("Decision Tree")
plt.show()

Output:-
 EXPERIMENT-7

Aim:- Implementation of K-means clustering.

Software Used:- Python

Theory:-K-means clustering is a widely used unsupervised machine learning algorithm for dividing a
dataset into distinct, non-overlapping groups or clusters. The algorithm's goal is to group similar data
points closely while keeping different clusters as distinct as possible. It works through an iterative
process, adjusting cluster centroids (representative points) until the solution converges. In Business
Intelligence (BI), K-means clustering has several valuable applications:

1. Customer Segmentation: Companies apply K-means clustering to group customers based on

similar purchasing behaviors, demographics, or preferences. This supports targeted
marketing, personalized offers, and enhanced customer satisfaction.
2. Anomaly Detection: K-means can spot outliers or anomalies in data, assisting businesses
in identifying irregularities or potential fraud in transactions, network activity, or
operational processes.
3. Product Categorization: In e-commerce and inventory management, K-means is used to group
products based on attributes like sales trends, demand, or other features. This helps in
optimizing inventory, pricing, and supply chain strategies.
4. Pattern Recognition: K-means can reveal patterns or trends within large datasets, helping BI
analysts understand complex data relationships and enabling data-driven decision-making
and strategic planning.
5. Document Clustering: In text analytics, K-means can categorize documents or articles by
content similarity, aiding in content categorization, topic modeling, and information
retrieval within BI systems.

Procedure:-
import numpy as np
import matplotlib.pyplot as plt
from sklearn import datasets
from sklearn.cluster import KMeans

# Load the Iris dataset

iris = datasets.load_iris()
X = iris.data

# Define the number of clusters (k)

k=3

# Initialize K-means using sklearn

kmeans = KMeans(n_clusters=k, random_state=0)
kmeans.fit(X)

# Get cluster assignments for each data

point cluster_assignments = kmeans.labels_

# Get cluster centers

cluster_centers = kmeans.cluster_centers_

# Visualize the results

colors = ['r', 'g', 'b']
markers = ['o', 's', 'd']
plt.figure(figsize=(10, 6))

for i in range(k):
plt.scatter(X[cluster_assignments == i, 0], X[cluster_assignments == i, 1],
color=colors[i], marker=markers[i], label=f'Cluster {i + 1}')

# Plot cluster centers

plt.scatter(cluster_centers[:, 0], cluster_centers[:, 1], c='k', marker='x', s=100, label='Centroids')

plt.xlabel('Feature 1')
plt.ylabel('Feature 2')
plt.legend()
plt.title('K-means Clustering (k=3)')
plt.show()

Output:-
Plot the clusters and their centers.
 EXPERIMENT -8

Aim: Implementation of Business Intelligence in Employee Productivity Tracker

Software used: Python

Theory:
Tracking employee productivity is essential for optimizing workplace efficiency and identifying areas of
improvement. Business Intelligence (BI) systems can analyse employee performance data to extract
meaningful insights and trends. This experiment demonstrates the use of Python to visualize and analyse
productivity metrics such as task completion, working hours, and efficiency.

Applications in BI:

1. Performance Analysis: Identify high-performing employees and areas where support is needed.
2. Task Prioritization: Analyse which tasks consume the most time to optimize workflows.
3. Resource Allocation: Ensure resources are distributed effectively to maximize productivity.
4. Employee Engagement: Understand patterns in engagement and motivation based on productivity
trends.

Procedure:
# Importing required libraries
import pandas as pd
import matplotlib.pyplot as plt
import seaborn as sns
# Sample dataset: Employee productivity data
data = {
'Employee': ['Alice', 'Bob', 'Charlie', 'David', 'Eve'],
'Hours Worked': [40, 35, 45, 50, 38],
'Tasks Completed': [10, 8, 12, 15, 9],
'Efficiency (%)': [90, 85, 92, 88, 87]
}
# Creating a DataFrame
df = pd.DataFrame(data)
# Displaying the dataset
print("Employee Productivity Data:")
print(df)

# Visualizing the data: Bar chart for Hours Worked vs Tasks Completed
plt.figure(figsize=(10, 6))
plt.bar(df['Employee'], df['Hours Worked'], color='blue', alpha=0.7, label='Hours Worked')
plt.bar(df['Employee'], df['Tasks Completed'], color='green', alpha=0.7, label='Tasks Completed')
plt.xlabel('Employee')
plt.ylabel('Count')
plt.title('Hours Worked vs Tasks Completed')
plt.legend()
plt.show()
# Scatter plot: Hours Worked vs Efficiency
plt.figure(figsize=(8, 5))
sns.scatterplot(data=df, x='Hours Worked', y='Efficiency (%)', hue='Employee', s=100)
plt.title('Employee Efficiency Analysis')
plt.xlabel('Hours Worked')
plt.ylabel('Efficiency (%)')
plt.show()
# Heatmap: Correlation analysis
plt.figure(figsize=(6, 4))
correlation = df[['Hours Worked', 'Tasks Completed', 'Efficiency (%)']].corr()
sns.heatmap(correlation, annot=True, cmap='coolwarm')
plt.title('Correlation Analysis')
plt.show()
# Insights: Identifying the most and least productive employee
most_productive = df.loc[df['Efficiency (%)'].idxmax()]
least_productive = df.loc[df['Efficiency (%)'].idxmin()]
print(f"Most Productive Employee: {most_productive['Employee']} (Efficiency: {most_productive['Efficiency
(%)']}%)")
print(f"Least Productive Employee: {least_productive['Employee']} (Efficiency: {least_productive['Efficiency
(%)']}%)")

Output: