Sourcing

1.​ Which data sources are already available and in use?

We have taken into account various sources, including Edmunds, Kaggle

repositories, providing detailed information on vehicle characteristics and

performance metrics.

2.​ Which internal data sources are available but not yet in use?

Internal sources of data that are not fully exploited in the automotive sector

include the significant telematics data streams from OEMs that are not used for

predictive analytics and operational optimization.

3.​ Which external data sources of partners or customers can be accessed?

Data from dealerships, insurance companies, parts suppliers, or even customer

usage data

4.​ Which public data sources can be accessed?

Government fuel efficiency data (EPA), safety ratings (NHTSA, IIHS), automotive

news and review sites, market reports.

5.​ Which additional data sources are required to enable the desired utilization?

Economic data (fuel prices, interest rates), demographics, competitor data, social

media sentiment.

6.​ Which (additional) data sources and data arise because of the key activities as

part of the new or changed business model?

Data from new online sales platforms, customer interactions on social media,

data from test drives, usage data from connected cars.

improved offering, which leads to more customers seizing the offer and in turn

produce more data?5

Analyzing user preferences can lead to better car designs, which attract more

customers, generating more data. Personalized recommendations can increase

sales, leading to more data on customer choices.

Refinement

1.​ Data cleansing & normalization:

Yes, handle missing values, correct errors, standardize formats, convert

categorical variables, and scale numerical variables.7

2.​ Data aggregation, integration & transformation

Combine data from different sources, calculate new features (e.g.,

horsepower-to-weight ratio), and transform data into suitable formats for analysis.

3.​ Data sampling or data compression

facilitating efficient analysis and model training. Meanwhile, data compression

techniques, such as dimensionality reduction and encoding, can optimize storage

and processing speed while preserving essential information.

4.​ Feature engineering & selection:

Create new features from existing ones (e.g., combining engine size and

horsepower). Select the most relevant features for your analysis.

Use ETL tools or scripting to load the cleaned and transformed data into a

database or data warehouse for analysis.

Utilization

1.​ Which value propositions for our customers do we want to create from the data?

Personalized car recommendations, price predictions, performance comparisons,

fuel efficiency estimations, identifying cars that best meet their needs.

2.​ How can the data improve our customer relationship?

Understanding customer preferences allows for targeted marketing and better

customer service.

3.​ How can the data make our (marketing, sales and distribution) channels more

efficient?

enables targeted marketing campaigns and optimized inventory management by

providing insights into consumer preferences and regional demand trends.

Additionally, it enhances sales strategies and improves distribution logistics,

ensuring that vehicles align with market needs and reducing operational costs.

4.​ How can we use the data to increase revenue?

Pricing optimization, product development, identifying new market segments.

Understanding customer needs, preferences, and pain points. Identifying trends

in the market.

6.​ How can we monetize our data in order to create new sources of revenue (e.g.,

data-as-a-service)?

Selling market research reports, providing API access to data and analytics,

personalized car recommendation services.

7.​ Which metrics and key performance indicators (KPI) can we calculate?

Average price by make/model, fuel efficiency, horsepower-to-weight ratio,

customer satisfaction (if feedback data is available), market share (if sales data is

available).

8.​ Which models for making predictions or recommendations can we create?

Price prediction, fuel efficiency prediction, car recommendation systems, sales

forecasting.

9.​ Which visualizations do we need for showing key figures, trends or correlations?

Scatter plots (relationship between features), bar charts (comparison of models),

histograms (distribution of features), line charts (trends over time).

10.​How do we integrate the analytics results into existing business processes?

Use insights to inform marketing campaigns, product development, pricing

strategies, and customer service improvements.

11.​Which manual steps are required in order to ensure quality throughout the

analytics process?

Data cleaning, feature engineering, model validation, and regular monitoring of

data and models.

ETL tools (Apache NiFi, Talend), scripting (Python), API connectors.

13.​Which database systems do we need for storing the data?

PostgreSQL, MySQL, or cloud-based databases (AWS RDS, Azure SQL).

14.​Which systems do we need for aggregating, integrating, and transforming the

data?

Python (pandas, NumPy), Apache Spark (for large datasets), dbt.

15.​Which analytical tools do we need for performing descriptive and diagnostic

analytics?

Power BI, Tableau, R, Python (SciPy, Statsmodels).

16.​Which modeling tools do we need for performing predictive and prescriptive

analytics?

Scikit-learn, TensorFlow, PyTorch.

People

1. Who takes care of data quality?

Data Steward: Ensuring data consistency (e.g., horsepower units are consistent).

○​ Completeness (handling missing values for features like MPG or engine

size).
2. Who is responsible for data protection?

●​ Data Protection Officer (DPO)

If the dataset includes any customer-related information (e.g., from sales records

or surveys), a DPO is essential.

○​ Ensure compliance with privacy regulations (GDPR, CCPA, etc.).

3. Who manages the IT systems?

●​ IT Administrator/Data Engineer:

○​ Maintaining the database or data warehouse where the automobile data is

stored.

○​ Managing access control to the data.

4. Who designs the analytics processes?

●​ Data Engineer/Data Architect: Designing the data pipeline for ingesting and

processing the automobile data.

○​ Integrating data from different sources (e.g., combining specifications with

sales data).

5. Who carries out the analysis?

●​ Data Scientist/Analyst: Exploring the data to identify trends and relationships.

○​ Performing statistical analysis on car specifications.

○​ Building predictive models (e.g., price prediction, fuel efficiency

prediction).
6. Who implements the analytics processes?

●​ Software Engineer/Data Engineer: Implementing the data processing and

analysis workflows.

○​ Developing scripts or programs for data manipulation and modeling.

7. Who interprets the results?

●​ Business Analyst/Domain Expert:

Translating the technical findings of the data analysis into business

recommendations.

8. Who creates and validates the prediction and recommendation models?

●​ Machine Learning Engineer/Data Scientist, Fine-tuning models to optimize their

effectiveness, Deploying models for real-world use.

9. Who designs the visualizations?

●​ Data Visualization Specialist/BI Developer: This role is essential for:

○​ Creating dashboards and reports that effectively communicate the insights

from the data analysis.

○​ Designing visualizations that are clear, concise, and engaging.

○​ Using tools like Power BI, Tableau, or Python libraries (Matplotlib,

Seaborn) to create visualizations. They would consider what information

would be best shown in a scatter plot vs. a bar chart, etc., for maximum

understanding.
Partners

1.​ Which partners or customers provide us with their data sources (data-on-demand

or data-as-a-service)?

Market research firms, automotive websites, dealerships (potentially).

2.​ Which product providers supply us with the necessary tools (as

software-on-demand or software-as-a-service)?

Cloud providers (AWS, Azure, Google Cloud), BI tool vendors (Microsoft,

Tableau), data science platform providers (Databricks).

3.​ Which implementation tasks can we outsource to external service providers?

Data cleaning, model development, dashboard creation.

4.​ Which responsibilities (roles) can we outsource to external service providers?

Data engineers, data scientists, BI developers.

5.​ Which external expert know-how do we need (consulting and/or training)?

Data science consulting, automotive industry expertise, cloud computing

consulting.