Improving Customer Value Index and Consumption Forecasts using a Weighted RFM Model and Machine Learning Algorithms

Collecting and mining customer consumption data are crucial to assess customer value and predict customer consumption behaviors. This paper proposes a new procedure, based on an improved Random Forest Model by: adding a new indicator, joining the RFMS-based method to a K-means algorithm with the Entropy Weight Method applied in computing the customer value index, classifying customers to different categories, and then constructing a consumption forecasting model whose RMSE is the smallest in all kinds of data mining models. The results show that identifying customers by this improved RMF model and customer value index facilitates customer profiling, and forecasting customer consumption enables the development of more precise marketing strategies.

Improving Customer Value Index and Consumption Forecasts Using a Weighted RFM Model and Machine Learning Algorithms

Collecting and mining customer consumption data are crucial to assess customer value and predict customer consumption behaviors. This paper proposes a new procedure, based on an improved Random Forest Model by: adding a new indicator, joining the RFMS-based method to a K-means algorithm with the Entropy Weight Method applied in computing the customer value index, classifying customers to different categories, and then constructing a consumption forecasting model whose RMSE is the smallest in all kinds of data mining models. The results show that identifying customers by this improved RMF model and customer value index facilitates customer profiling, and forecasting customer consumption enables the development of more precise marketing strategies.

Prediction of Domestic Hot Water Temperature in a District Heating Network

With the fourth generation of district heating networks in sight, opportunities are rising for better services and optimized planning of energy production. Indeed, the more intensive data collection is expected to allow for load prediction, customer profiling, etc. In this context, our work aims at a better understanding of customer profiles from the captured data. Given the variety of households, such profiles are difficult to capture. This study explores the possibility to predict domestic hot water (DHW) usage. Such prediction is made challenging due to the presence of two components in the signal, the first one bound to the physical properties of the DHW distribution system, the second one bound to the human patterns related to DHW consumption. Our contributions include (1) the analysis of recurrent neural network architectures based on GRU, (2) the inclusion of state-based labels inferred in an unsupervised way to simulate domain knowledge, (3) the comparison of different features. Results show that the physical contribution in the signal can be forecasted successfully across households. On the contrary, the stochastic “human” component is harder to predict and would need further research, either by improving the modelling or by including alternate signals.

Security of the Use of Cognitive Information Technologies of Decision-Making

The article considers the intelligent automation of decision-making and management procedures that is being implemented in many areas of socio-economic practice, including financial and credit business processes, in trade and e-commerce (customer profiling, marketing micro-targeting), telecommunications, industry (technological control, robotics, neurocontrol, strategic planning and forecasting), intelligent automation also came to business management, to public administration. It is claimed that automation of personnel management is expanding (monitoring compliance with requirements, profiling and assessing KPIs, predicting conflicts and violations), unmanned vehicles and other neural network automation are used in medicine, the transport industry and agriculture; smart technologies come to education (in Moscow, a system of predictive analytics of the digital footprint of students is being tested to optimize and target educational services, help orientate in the future profession). The use of cognitive technologies in the creation of expert, advisory systems, decision support systems provides not only convenience and savings in time and effort, but gives rise to a variety of organizational, economic, ethical, social problems, giving rise to new risks. This study provides an overview of intelligent technologies that are used in social management, threats associated with the practical use of intelligent automation tools and decision support, ways and measures to reduce some of the risks associated with these threats.

Customer profile analysis of the mail-order pharmacy MedicineGo using the example of the German mail-order pharmacy MedicineGo

Pharmacies in transition - digitalisation measures, connection to the telematics infrastructure and the oligopoly position of leading mail-order pharmacies. Not only meeting the technical implementations in the digital transformation, but also satisfying customers are important issues for public pharmacies. While retail pharmacies have access to data on their regular customers, we know little about customers of mail-order pharmacies. As the trend to order medicines online grows, mail-order medicines will take on an increasingly important role. It is expected that the planned introduction of electronic prescriptions in Germany in mid-2021 will lead to a further increase in online sales. The research question is: What does the customer profile of a mail-order pharmacy look like? Thus, this study focuses on an economic customer profiling of a mail-order pharmacy using the example of the nationally operating mail-order pharmacy MedicineGo. Based on the assumption that younger people between 18 and 50 years of age prefer to use the internet as an ordering medium, due to a more savvy use of computers, a quantitative analysis of empirical data will present essential market research data. In order to interpret the results of the quantitative analysis and to establish causalities, additional expert interviews with the MedicineGo administrator and other employees of the mail-order pharmacy as well as a qualitative literature analysis were conducted by means of qualitative analysis. By means of quantitative analysis, it was empirically shown that with regard to the customer profile of the mail-order pharmacy MedicineGo, 72% of customers use the direct web shop as a sales channel for their orders, the proportion of women is about ten percentage points higher than that of men, and the largest purchasing groups by number of orders are persons between 51 and 80 years of age. The quantitative analysis showed that older people (over 50 years) prefer to use the mail-order pharmacy, so another result of the study is that public pharmacies compete with mail-order pharmacies for the same customer group.

Segmentasi Pelanggan Produk Digital Service Indihome Menggunakan Algoritma K-Means Berbasis Python

PT. Telekomunikasi Indonesia adalah salah satu perusahaan yang mengedepankanpelanggan akan tetapi belum ada informasi tentang karakteristik pelanggan. Pada penelitian ini dilakukan analisa karakteristik pelanggan sebagai dasar penetapan segmentasi pelanggan dan customer profiling pelanggan produk digital service add on Indihome menggunakan Algoritma K-Means. Penentuan jumlah cluster terbaik dilakukan menggunakan metode Elbow dan diperoleh nilai K = 3, sehingga data pelanggan dikelompokkan kedalam tiga segmen. Pengolahan data pelanggan dibagi menjadi 3 simulasi dengan persentase data train dan data test 80% - 20%, 70% - 30% dan 50% - 50%. Data yang digunakan berjumlah 1392 record sebagai populasi imanadata tersebut akan digunakan untuk mencari karakteristik setiap data tersebut. Evaluasi cluster dilakukan menggunakan metode Silhouette Index, Davies Bouldin Index dan Calinski Harabasz Index. Hasil dari penelitian menunjukan bahwa simulasi ketiga merupakan simulasi terbaik berdasarka evaluasi cluster dengan presentasi data train 50% dan data test 50% dimana customer profiling dilihat dengan menganalisis anggota masing-masing cluster dari simulasi ketiga dimana cluster 0 memiliki anggota 396 pelanggan dengan kategori pelanggan yang memberikan keuntungan terbesar bagi perusahaan, cluster 1 memiliki anggota 286 pelanggan dengan kategori pelanggan yang tanpa disadari memiliki potensi besar dalam memberikan keuntungan bagi perusahaan, dan cluster 2 memiliki anggota 14 pelanggan dengan kategori pelanggan yang memberikan keuntungan lebih sedikit daripada biaya untuk memberikan pelayanan.