This Thesis focuses on the retail sector and aims to design and implement a price optimization system that suggests the prices to be applied to a company’s products, in order to maximize profit. Currently, the considered company sets the prices aligning them with the average market price, while the approach proposed in this Thesis exploits the information deriving from the transactional company data in order to forecast the sales of the products and use this information to suggest prices. A first phase of the work includes the exploration of the company’s historical data, aimed at characterizing the pricing strategy currently adopted, through the use of data analysis tools. The second phase consists in the implementation of the price optimization system, which includes a data preprocessing phase, a sales prediction phase and a price prescription phase. An original contribution of this work consists in a data pre-processing technique that improves the accuracy of the sales prediction, also limiting the execution time. This technique provides for the definition of an appropriate measure of similarity between different shops, and on the basis of it determines groups of shops that are homogeneous in terms of the pricing strategy adopted. The technique is general as it is based on transactional data and not on descriptive information about the shops. The sales prediction phase takes into account both time factors and product characteristics. The developed system has been tested experimentally in different business scenarios. The results are discussed both qualitatively and quantitatively, comparing the expected performance of the system with those achieved with the pricing strategy adopted by the company.
Questa Tesi si colloca nell’ambito della Grande Distribuzione Organizzata e si prefigge di progettare e implementare un sistema che suggerisca i prezzi da applicare sui prodotti di una azienda, allo scopo di massimizzare il profitto. Attualmente, l’azienda in questione stabilisce i prezzi allineandoli al prezzo medio di mercato, mentre l’approccio proposto in questa Tesi sfrutta l’informazione derivante dai dati transazionali aziendali al fine di prevedere le vendite dei prodotti e usare tale informazione per suggerirne i prezzi. Una prima fase del lavoro comprende l’esplorazione dei dati storici dell’azienda, volta a caratterizzare la strategia di pricing correntemente adottata, mediante l’uso di strumenti dell’analisi dei dati. La seconda fase consiste nell’implementazione del sistema di ottimizzazione del prezzo, che comprende una fase di preprocessazione dei dati, una fase di predizione delle vendite e una di prescrizione dei prezzi. Un’originale contributo di questo lavoro consiste in una tecnica di preprocessazione dei dati che migliora l’accuratezza della predizione delle vendite, limitandone inoltre il tempo di esecuzione. Tale tecnica prevede la defizione di una appropriata misura di similarità tra diversi negozi, e sulla base di essa determina gruppi di negozi omogenei in termini di strategia di pricing adottata. La tecnica è generale in quanto si basa su dati transazionali e non su informazioni descrittive dei negozi. La fase di predizione delle vendite tiene conto sia di fattori temporali, sia relativi alle caratteristiche del prodotto. Il sistema sviluppato è stato testato sperimentalmente in diversi scenari di business. I risultati sono stati discussi sia qualitativamente che quantitativamente, comparando le prestazioni attese del sistema con quelle raggiunte con la strategia di pricing adottata dall’azienda.
Design and test of a data-driven methodology for suggesting the optimal pricing strategy in the retail industry
De MARINI, BRUNO
2018/2019
Abstract
This Thesis focuses on the retail sector and aims to design and implement a price optimization system that suggests the prices to be applied to a company’s products, in order to maximize profit. Currently, the considered company sets the prices aligning them with the average market price, while the approach proposed in this Thesis exploits the information deriving from the transactional company data in order to forecast the sales of the products and use this information to suggest prices. A first phase of the work includes the exploration of the company’s historical data, aimed at characterizing the pricing strategy currently adopted, through the use of data analysis tools. The second phase consists in the implementation of the price optimization system, which includes a data preprocessing phase, a sales prediction phase and a price prescription phase. An original contribution of this work consists in a data pre-processing technique that improves the accuracy of the sales prediction, also limiting the execution time. This technique provides for the definition of an appropriate measure of similarity between different shops, and on the basis of it determines groups of shops that are homogeneous in terms of the pricing strategy adopted. The technique is general as it is based on transactional data and not on descriptive information about the shops. The sales prediction phase takes into account both time factors and product characteristics. The developed system has been tested experimentally in different business scenarios. The results are discussed both qualitatively and quantitatively, comparing the expected performance of the system with those achieved with the pricing strategy adopted by the company.File | Dimensione | Formato | |
---|---|---|---|
2019_07_DeMarini.PDF
non accessibile
Descrizione: Testo della tesi
Dimensione
3.89 MB
Formato
Adobe PDF
|
3.89 MB | Adobe PDF | Visualizza/Apri |
I documenti in POLITesi sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.
https://hdl.handle.net/10589/148599