Description
In the rapidly evolving telecommunications industry, the focus on predictive ana-lytics has grown significantly with the popularization of machine learning models.
These advancements have proven crucial for optimizing business strategies, mar-
keting efforts, and financial decision-making. This thesis explores how predictive
analytics can address challenges faced by telecommunications companies in the B2B
sector, particularly in identifying high-potential customers within a broad and di-
verse market. By integrating machine learning models with customer segmentation
techniques, this research aims to enhance the accuracy of revenue prediction, en-
abling more targeted marketing efforts and strategic decision-making while mini-
mizing business risks.
A key challenge in applying predictive analytics is the limited availability of time-
series data, which often fails to capture the full scope of customer behavior across
the broader market. To overcome this, the thesis utilizes cross-sectional snapshot
data, enriched with telecom billing information, offering a more inclusive view of
customer attributes such as demographics, industry type, and revenue size.
The thesis proposes a segmentation and machine learning framework developed
for Telia, a Finnish telecommunications company. The framework applies DBSCAN
and K-Means clustering algorithms respectively to identify meaningful customer pat-
terns. A random forest model is trained on high-quantile data of each cluster, cap-
turing the characteristics of high-value business customers across different segments
to forecast a potential revenue.
The findings reveal that the predictive framework performs similarly to the base-
line cluster mean, slightly outperforming it for medium to low-revenue customers
but lagging behind on high-revenue customers. While the framework is effective
for identifying lower-value customers, further refinement may be necessary to better
capture the complexities of high-end customer behavior. However, the integration
of these methodologies creates a complementary approach, which offers valuable in-
sights that could be leveraged to enhance targeting strategies across all customer
segments, ultimately leading to more effective decision-making and optimized busi-
ness practices in the telecommunications industry.
| Period | 20 Mar 2025 |
|---|---|
| Examinee | |
| Degree of Recognition | National |
Country of activity
- Finland