Garuda - Garba Rujukan Digital

Segmenting Walmart Customers for Personalized Marketing Strategies Using MiniBatchKMeans Clustering and Decision Trees: An Analysis of Purchasing Behavior Buchdadi, Agung Dharmawan
Journal of Digital Market and Digital Currency Vol. 1 No. 3 (2024): Regular Issue December
Publisher : Bright Publisher

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.47738/jdmdc.v1i3.18

This study explores the application of MiniBatchKMeans clustering and decision tree analysis to segment Walmart customers for personalized marketing strategies. Using a dataset of 550,068 customer transactions, including variables such as User_ID, Product_ID, Gender, Age, Occupation, City_Category, Stay_In_Current_City_Years, Marital_Status, Product_Category, and Purchase, we identified five distinct customer segments. These segments were characterized by unique demographic and purchasing behaviors. Segment 1 included older customers (mean age: 55+) with high and consistent spending, primarily on premium products. Segment 2 comprised middle-aged customers (mean age: 36-45) with moderate to high spending levels, favoring household and family-related products. Segment 3 consisted of young adults (mean age: 18-25) with variable purchasing patterns, focusing on low to mid-range priced items. Segment 4 included young families (mean age: 26-35) with significant spending on a variety of products, and Segment 5 featured middle-aged to older customers (mean age: 46-55) with steady but moderate spending habits. The MiniBatchKMeans clustering algorithm effectively handled the large dataset, identifying clear customer segments. Decision tree analysis provided insights into the key features driving each segment, with Purchase amount, Age, and Occupation being the most significant. The decision tree model achieved an accuracy of 100%, with precision, recall, and f1-scores of 1.00 for all segments, indicating robust classification. These findings have significant implications for personalized marketing strategies. For instance, premium product promotions can be directed at high-spending older customers, while family-oriented discounts and bundles can be tailored for young families. Digital marketing efforts can be optimized to engage younger segments through social media and personalized recommendations. This study highlights the importance of data-driven decision-making in retail, emphasizing the need for continuous data collection and analysis to stay competitive. Future research should incorporate datasets from different retail contexts and explore alternative clustering techniques and additional features to provide a more holistic view of customer segmentation.

Optimizing Publisher Revenue in Digital Marketing Using Decision Trees and Random Forests Irfan, Muhamad
Journal of Digital Market and Digital Currency Vol. 1 No. 3 (2024): Regular Issue December
Publisher : Bright Publisher

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.47738/jdmdc.v1i3.19

This study explores the optimization of reserve prices in real-time first price auctions within digital advertising using decision tree and random forest algorithms. The dataset used includes 567,291 entries covering various variables such as impressions, bids, prices, and revenue, providing a comprehensive view of auction dynamics over a full year. The decision tree model achieved a Mean Squared Error (MSE) of 0.1347 and an R² score of 0.731, indicating a reasonable level of accuracy in predicting reserve prices. In contrast, the random forest model significantly outperformed the decision tree model with an MSE of 0.0789 and an R² score of 0.842, demonstrating superior predictive power and robustness. The analysis revealed that the application of these machine learning models significantly enhances the accuracy and reliability of reserve price predictions, helping publishers to optimize their revenue. The findings show that by setting optimal reserve prices based on the models' predictions, publishers can minimize the risk of underselling ad inventory and maximize revenue, as evidenced by a 15% increase in revenue observed in a case study after implementing the random forest model. The study also provides insights into bidder behavior, particularly bid shading strategies, highlighting how bidders adjust their bids in response to different reserve price settings. Higher reserve prices tend to reduce bid shading, resulting in more competitive and balanced auctions. The practical implications for digital marketing include enhanced strategic decision-making for publishers and a more transparent and predictable bidding environment for advertisers. Despite the promising results, the study acknowledges limitations such as reliance on historical data from a single ad exchange platform and the assumptions inherent in the models. Future research should expand the dataset to include multiple platforms and explore more advanced machine learning techniques to further improve reserve price optimization. Overall, this research underscores the potential of leveraging data science and machine learning to transform digital advertising strategies, driving higher revenue and efficiency in the industry.

Predicting Ad Click-Through Rates in Digital Marketing with Support Vector Machines Sangsawang, Thosporn
Journal of Digital Market and Digital Currency Vol. 1 No. 3 (2024): Regular Issue December
Publisher : Bright Publisher

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.47738/jdmdc.v1i3.20

This study investigates the effectiveness of Support Vector Machines (SVM) in predicting click-through rates (CTR) in digital marketing campaigns. Utilizing a dataset comprising user demographic and behavioral data, the research aims to develop a predictive model to forecast ad clicks accurately. The primary objectives include conducting exploratory data analysis (EDA), preprocessing data, training the SVM model, and evaluating its performance using standard metrics. The dataset includes features such as Daily Time Spent on Site, Age, Area Income, Daily Internet Usage, and Gender. Key findings from the EDA reveal that "Daily Time Spent on Site" and "Daily Internet Usage" are significant predictors of CTR, with notable correlations. The SVM model, trained on this data, demonstrated exceptional performance, achieving an accuracy of 97.65%, a precision of 98.58%, a recall of 96.53%, and an F1-score of 97.54%. These results confirm the model's robustness and reliability, indicating its potential for optimizing digital marketing strategies. The study's significance lies in its contribution to the fields of digital marketing and predictive analytics by showcasing the applicability and advantages of SVM in predicting user behavior. These insights can help marketers optimize ad placements, enhance user engagement, and improve return on investment. Practical implications include strategies for targeted and personalized marketing based on key user demographics and behaviors. Despite the promising results, the study acknowledges limitations such as the dataset size and scope of features. Future research should focus on utilizing larger and more diverse datasets, incorporating additional features, and exploring other advanced machine learning algorithms. This research encourages further exploration of machine learning applications in digital marketing to enhance predictive accuracy and campaign effectiveness. By addressing these aspects, this study aims to advance the academic understanding and practical implementation of predictive analytics in digital marketing, providing a robust framework for future applications.

Analyzing the Determinants of User Satisfaction and Continuous Usage Intention for Digital Banking Platform in Indonesia: A Structural Equation Modeling Approach Pratama, Satrya Fajri
Journal of Digital Market and Digital Currency Vol. 1 No. 3 (2024): Regular Issue December
Publisher : Bright Publisher

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.47738/jdmdc.v1i3.21

This study investigates the factors influencing user satisfaction (US) and continuous usage intention (UI) of the digital banking platform in Indonesia. Utilizing a quantitative research approach, structural equation modeling (SEM) via SmartPLS was employed to analyze data from 376 users. The study integrates key constructs, including Task-Technology Fit (TTF), System Quality (SQ), Performance Expectancy (PE), US, and UI, into a comprehensive model. The findings confirm that TTF, SQ, PE, and US significantly influence UI. Specifically, higher TTF and SQ directly enhance PE (path coefficient = 0.871, t-value = 92.895) and US (path coefficient = 0.798, t-value = 47.957), positively impacting UI. Performance Expectancy emerged as a stronger predictor of UI (path coefficient = 0.559, t-value = 12.800) compared to the US (path coefficient = 0.245, t-value = 5.229), underscoring the critical role of perceived performance benefits in driving continuous usage. All five hypotheses were supported: TTF positively affects UI (path coefficient = 0.250, t-value = 7.154); SQ positively influences PE and US; PE positively impacts UI; and US positively affects UI. The Sobel test results indicated that PE significantly mediates the relationship between SQ and UI (Z = 12.60), and US also significantly mediates this (Z = 5.19). The R-squared values indicate the explanatory power of the model: PE (0.758), UI (0.956), and US (0.637), demonstrating that the model explains a substantial portion of the variance in these constructs. The study contributes to the literature by validating the integrated model, extending existing models such as the Technology Acceptance Model (TAM) and the Unified Theory of Acceptance and Use of Technology (UTAUT), and highlighting the importance of technical and perceptual factors in technology adoption. Practically, the results offer actionable insights for digital banking providers. Enhancing TTF and maintaining high SQ are crucial for fostering positive user experiences and encouraging continuous usage. Providers should also emphasize the performance benefits of their platforms to improve PE and UI. Despite its contributions, the study has limitations, including sample size and reliance on self-reported data, which may affect generalizability. Future research could expand the sample size, incorporate objective usage data, and explore additional factors such as social influence and facilitating conditions. Overall, the study provides a robust framework for understanding user behavior in digital banking and offers practical strategies for improving user satisfaction and retention in the industry.

Using K-Means Clustering to Enhance Digital Marketing with Flight Ticket Search Patterns Sukmana, Husni Teja; Oh, Lee Kyung
Journal of Digital Market and Digital Currency Vol. 1 No. 3 (2024): Regular Issue December
Publisher : Bright Publisher

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.47738/jdmdc.v1i3.22

This study explores the application of K-Means clustering to enhance digital marketing strategies by analyzing flight ticket search patterns. Utilizing a dataset containing 4,000 search engine results related to flights to Hong Kong, the research identifies five distinct user clusters based on search terms, titles, snippets, and other relevant features. The dataset's key features include search terms, ranks, titles, snippets, display links, and direct links, providing a comprehensive view of user interactions and preferences. The cluster analysis reveals significant variations in user intent and preferences across the identified segments. For instance, Cluster 1 is characterized by users searching for "cheap flights" and "discount tickets," indicating a price-sensitive segment. In contrast, Cluster 2 users prefer "premium flights" and "business class," highlighting an interest in luxury travel options. The study also examines the behavioral patterns within each cluster, such as Cluster 3 users who search for flights well in advance and prioritize flexible booking options. The findings underscore the effectiveness of K-Means clustering in enhancing digital marketing strategies. By leveraging the insights from the clustering analysis, marketers can design highly targeted advertising campaigns and personalized offers. For example, budget airlines can target Cluster 1 with promotions and discounts, while premium airlines can focus on Cluster 2 with exclusive service highlights. This targeted approach is expected to improve user engagement and conversion rates significantly. The study also highlights the advantages of behavior-based segmentation over traditional demographic methods, offering a more accurate representation of user preferences and intentions. The identified clusters provide a framework for understanding different user groups, enabling more efficient resource allocation and campaign design. Future research should explore the integration of additional data sources, such as social media interactions and user reviews, to enhance clustering accuracy. Additionally, advanced clustering techniques like hierarchical clustering and Gaussian Mixture Models could be investigated to provide further insights. The ongoing refinement and enhancement of segmentation processes are crucial for maintaining effective and impactful digital marketing strategies in the dynamic travel industry. Key results include the identification of five user clusters, the importance of personalized marketing strategies, and the potential for improved engagement and conversion rates through targeted advertising and offers.

Volatility and Risk Assessment of Blockchain Cryptocurrencies Using GARCH Modeling: An Analytical Study on Dogecoin, Polygon, and Solana Doan, Minh Luan
Journal of Digital Market and Digital Currency Vol. 2 No. 1 (2025): Regular Issue March
Publisher : Bright Publisher

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.47738/jdmdc.v2i1.25

This study analyzed the volatility and risk profiles of three prominent blockchain-based cryptocurrencies—Dogecoin, Polygon, and Solana—using the Generalized Autoregressive Conditional Heteroskedasticity (GARCH) model. Volatility, a key risk metric for cryptocurrencies, was modeled through the GARCH(1,1) framework, which effectively captured the time-varying nature of price fluctuations. The analysis revealed that Dogecoin exhibited the highest volatility and risk, primarily driven by its speculative market behavior and social media influence. Polygon and Solana, while also volatile, demonstrated more stability, with their risk profiles reflecting the technological advancements and broader use cases within their respective blockchain ecosystems. The study also incorporated Value at Risk (VaR) and Conditional Value at Risk (CVaR) metrics to assess the potential downside risks for each cryptocurrency. Dogecoin had the highest potential for extreme losses, followed by Polygon and Solana. The GARCH model successfully identified the volatility persistence in these assets, showing that past market conditions heavily influenced future volatility. This research contributes to the literature on cryptocurrency volatility by applying the GARCH(1,1) model to analyze digital assets with varying market characteristics. The findings emphasize the need for robust risk management strategies tailored to the unique behaviors of individual cryptocurrencies. Limitations of the study included the use of historical data and the focus on only three cryptocurrencies, suggesting opportunities for future research. Potential areas for further study include the incorporation of additional variables, such as macroeconomic indicators, and the exploration of alternative volatility models, such as EGARCH or TGARCH, to better capture the complexities of cryptocurrency markets. These insights provide valuable guidance for investors, risk managers, and policymakers navigating the volatile and evolving landscape of blockchain-based digital assets.

Enhancing Customer Satisfaction and Product Quality in E-commerce through Post-Purchase Analysis using Text Mining and Sentiment Analysis Techniques in Digital Marketing Izumi, Calvina; Ghaffar, Soeltan Abdul; Setiawan, Wilbert Clarence
Journal of Digital Market and Digital Currency Vol. 2 No. 1 (2025): Regular Issue March
Publisher : Bright Publisher

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.47738/jdmdc.v2i1.26

This study explores the application of text mining and sentiment analysis to enhance product quality and customer satisfaction within the e-commerce landscape. Using the Customer360Insights dataset, which comprises 236 records of customer interactions, demographic details, product information, and transactional data, we identified key drivers of negative feedback and returns. The descriptive statistics revealed a diverse customer base with an average age of 45.33 years and significant variability in monthly income ($5,470.24 ± $1,442.80). The text mining process, including tokenization and term frequency analysis, identified frequent terms such as "poor" (95 occurrences), "arrived" (92 occurrences), and "damaged" (45 occurrences). Sentiment analysis using VADER and TextBlob indicated that 80.08% of the feedback was negative, highlighting pervasive dissatisfaction. Topic modeling using Latent Dirichlet Allocation (LDA) revealed five main topics, consistently emphasizing issues like product quality and delivery timeliness. Common return reasons included poor value (55 occurrences), wrong item delivered (49 occurrences), and late arrivals (47 occurrences). These insights suggest critical areas for improvement, such as enhancing quality control, optimizing logistics, and refining pricing strategies. The findings have significant implications for digital marketing strategies, emphasizing the need for targeted interventions to improve customer satisfaction. By addressing identified issues and leveraging data-driven insights, e-commerce businesses can enhance their product offerings, optimize post-purchase support, and foster customer loyalty. Future research should validate these findings using real-world data and explore additional data mining techniques to provide a comprehensive understanding of customer satisfaction drivers.

Evaluating the Effectiveness of Digital Marketing Campaigns through Conversion Rates and Engagement Levels Using ANOVA and Chi-Square Tests Rahardja, Untung; Aini, Qurotul
Journal of Digital Market and Digital Currency Vol. 2 No. 1 (2025): Regular Issue March
Publisher : Bright Publisher

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.47738/jdmdc.v2i1.27

This study investigates the effectiveness of various digital marketing campaign types—Awareness, Conversion, and Retention—on conversion rates and engagement levels. Using a dataset of 8,000 records, we conducted a comprehensive analysis through ANOVA, Chi-Square tests, and OLS regression to understand the impact of these campaign types. The ANOVA results indicated no significant differences in conversion rates across the campaign types, with an F-statistic of 0.4752 and a p-value of 0.6218. Similarly, the analysis for engagement levels, measured by website visits, yielded an F-statistic of 0.3651 and a p-value of 0.6942, suggesting no significant differences among the campaigns. Despite these findings, the Chi-Square test revealed a significant association between campaign types and conversion outcomes, with a Chi-Square statistic of 84.4544 and a p-value of approximately 3.3983e-18. This suggests that while the overall conversion rates do not differ significantly, the type of campaign does influence whether conversions occur. Pairwise t-tests supported these results, showing no significant differences in conversion rates or engagement levels between specific pairs of campaign types. Further, OLS regression analysis for conversion rates resulted in an R-squared value of 0.001 and a non-significant F-statistic, indicating that the predictors such as AdSpend and ClickThroughRate do not significantly explain the variation in conversion rates. Similarly, the regression model for engagement levels, despite an R-squared value of 1.000, highlighted issues of multicollinearity and overfitting. These findings imply that simply altering the type of campaign may not substantially impact conversion rates or engagement levels. Marketers should focus on improving content quality, targeting precision, and user experience to enhance campaign effectiveness. Future research should incorporate additional variables and advanced modeling techniques to provide deeper insights into the factors driving digital marketing success.

Optimizing Pricing Strategies for Female Fashion Products Using Regression Analysis to Maximize Revenue and Profit in Digital Marketing Siddique, Quba
Journal of Digital Market and Digital Currency Vol. 2 No. 1 (2025): Regular Issue March
Publisher : Bright Publisher

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.47738/jdmdc.v2i1.28

This study explores optimal pricing strategies for the female fashion sector through the application of advanced data science methodologies. Utilizing a dataset of 4,272 entries, comprising various attributes such as original prices, promotional prices, and discount percentages, we employed regression models to predict promotional pricing. The research highlights Ridge Regression as the most effective model, balancing high accuracy with reduced overfitting. The model achieved an R-squared (R²) value of 0.9999999999999678, a Mean Absolute Error (MAE) of 4.31×10−6, and a Mean Squared Error (MSE) of 4.89×10 −11, demonstrating its robustness and reliability. The study's findings indicate that dynamic pricing and tailored discount strategies can significantly enhance revenue and profitability. High-value items are best priced with moderate discounts, maintaining higher promotional prices, while low-value items benefit from aggressive discounting to drive sales volume. Sensitivity analysis further supported these strategies by showing that a 10% increase in original prices proportionally increased promotional prices, while a 10% increase in discount percentages led to lower promotional prices, affecting sales performance differently across product categories. Practical implications for e-commerce businesses include implementing dynamic pricing, developing targeted discount strategies, and timing promotions strategically. Regular sensitivity analysis and continuous model validation are recommended to adapt to market changes effectively. Future research should consider broader datasets, advanced modeling techniques, external market factors, and customer segmentation to enhance the generalizability and applicability of pricing strategies across different sectors. This research underscores the importance of data-driven approaches in optimizing digital marketing strategies, offering actionable insights that can significantly boost revenue and profitability in the female fashion sector.

Predictive Analysis for Optimizing Targeted Marketing Campaigns in Bike-Sharing Systems Using Decision Trees, Random Forests, and Neural Networks Warmayana, I Gede Agus Krisna; Yamashita, Yuichiro; Oka, Nobuta
Journal of Digital Market and Digital Currency Vol. 2 No. 1 (2025): Regular Issue March
Publisher : Bright Publisher

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.47738/jdmdc.v2i1.29

This research explores the use of machine learning models to predict bike rental demand and optimize targeted marketing campaigns in bike-sharing systems. Utilizing the day.csv and hour.csv datasets, which provide daily and hourly bike rental data, we implemented Decision Tree Regressor, Random Forest Regressor, and Neural Networks (MLPRegressor) to forecast demand. The Random Forest model outperformed the others, achieving an RMSE of 709.08 and an MAE of 469.99 for daily predictions, while the Neural Network demonstrated potential for hourly forecasts. Our analysis revealed significant trends, including increased demand during summer months and peak usage times on weekday mornings and evenings, highlighting the importance of temporal and weather-related factors in predicting bike rental demand. The study's predictive insights allow bike-sharing companies to enhance operational efficiency by optimizing bike allocation during peak periods and reducing idle capacity during off-peak times. Furthermore, the ability to predict demand accurately enables the development of data-driven marketing strategies, such as launching promotions during high-demand periods and targeting specific user groups based on rental patterns. Despite the promising results, challenges such as data preprocessing complexities and computational resource constraints were encountered. Additionally, the study's scope was limited by the available data, suggesting the need for future research to incorporate additional data sources, like real-time traffic conditions and social events, and to explore more advanced machine learning techniques to further improve prediction accuracy. In conclusion, this research underscores the value of predictive analytics in optimizing bike-sharing systems and marketing strategies, contributing to more efficient and user-centric urban mobility solutions.

Home Page

OAI Link

Editorial Team

Contact

Reviewer

Google Scholar

Contact Name
Andhika Rafi Hananto

Contact Email
andhikarh90@gmail.com

Phone
+6282314736799

Journal Mail Official
support@jdmdc.com

Editorial Address
Graha Permata Estate, Jl. HM Bahrun Blok H9, Sokayasa, Berkoh, Kec. Purwokerto Tim., Kabupaten Banyumas, Jawa Tengah 53146

Location
Kab. banyumas,

Jawa tengah

INDONESIA

Abstract

Abstract

Abstract

Abstract

Abstract

Abstract

Abstract

Abstract

Abstract

Abstract

Home Page

OAI Link

Editorial Team

Contact

Reviewer

Google Scholar

Contact Name Andhika Rafi Hananto

Contact Email andhikarh90@gmail.com

Phone +6282314736799

Journal Mail Official support@jdmdc.com

Editorial Address Graha Permata Estate, Jl. HM Bahrun Blok H9, Sokayasa, Berkoh, Kec. Purwokerto Tim., Kabupaten Banyumas, Jawa Tengah 53146

Location Kab. banyumas, Jawa tengah INDONESIA

Abstract

Abstract

Abstract

Abstract

Abstract

Abstract

Abstract

Abstract

Abstract

Abstract

Contact Name
Andhika Rafi Hananto

Contact Email
andhikarh90@gmail.com

Phone
+6282314736799

Journal Mail Official
support@jdmdc.com

Editorial Address
Graha Permata Estate, Jl. HM Bahrun Blok H9, Sokayasa, Berkoh, Kec. Purwokerto Tim., Kabupaten Banyumas, Jawa Tengah 53146

Location
Kab. banyumas,

Jawa tengah

INDONESIA