Application Of Classification Trees Research Articles

Prediction of the Wine Price Purchased Using Classification Trees

The wine market is expected to recover and growth in the following years. Although the market size of this market is small in Mexico; efforts are being made to promote national wines. To support these actions, the current study aims to explore consumers’ decision to pay for a low versus high priced wine according to wine attributes and external influences. A total of 290 wine consumers, knowledgeable about wines, participated in an online survey that took place in the Northwest part of Mexico. The relevance that eleven attributes have on the wine price was assessed through the application of classification trees, a method that has not been previously applied to explore wine purchase decisions. The result indicate consumers are mainly driven by price than the wine attributes. However, relevant attributes for consumers of low-priced wines (less than 20 USD) include “label information” “awards” and “previous tasting”. By contrast, purchase decision of consumers of pricey wines are mainly driven by “country or region of origin” and “vineyard recognition”. The study findings offer provide suggestions on the wine attribute wineries and marketers should focus on to meet the current consumers’ preferences depending on the price-sensitivity and wine knowledge of target segments.

Comparisons of decision tree methods using water data

ABSTRACTThis article demonstrates the application of classification trees (decision trees), logistic regression (LR), and linear discriminant function (LDR) to classify data of water quality (i.e., whether the water is fit for drinking on not fit for drinking). The data on water quality were obtained from Pakistan Council of Research in Water Resources (PCRWR) for two cities of Pakistan—one representing industrial environment (Sialkot) and the other one representing non-industrial environment (Narowal). To classify data on water quality, three statistical tools were employed—the Decision Tree methodology using Gini Index, LR, and LDA—using R software library. The results obtained by the said three techniques were compared using misclassification rates (a model with minimum value of misclassification rate is better). It was witnessed that LR performed well than the other two techniques while the Decision trees and LDA performed equally well. But for illustration purposes decision trees technique is comparatively easy to draw and interpret.

Application of classification trees in determining the impact of phenotypic factors on conception to first service in Holstein cattle

The aim of the study was to evaluate the impact of phenotypic factors on the probability of conception to first service. The study analysed 49 317 lactation records gathered between 1995 and 2007 in Iran, detailing six consecutive lactations in 13 herds of Holstein cows. The impact of phenotypic factors on the conception to first service was evaluated using the classification tree technique. The CHAID algorithm with the Pearson Chi-squared test was used as the division criterion. The resulting tree was six levels deep and consisted of 22 leaves. The greatest divisions in the tree occurred based on the year of calving (five divisions), herd and parity (four divisions each), level of milk production and age at calving (three divisions each). Fewer divisions occurred based on the month when the first insemination was performed (two divisions). The tree structure obtained in the present study allowed identification of combinations of levels of factors associated with highest probabilities of conception. The highest rate of conception to first service was recorded for cows that: while calving were younger than 5 years old; were in herds 2, 4, 5, 7, 8, 12 and 14; and were inseminated in January, February, March, April, May and November. Those cows were at least in their third lactation.

A Method for Application of Classification Tree Models to Map Aquatic Vegetation Using Remotely Sensed Images from Different Sensors and Dates

In previous attempts to identify aquatic vegetation from remotely-sensed images using classification trees (CT), the images used to apply CT models to different times or locations necessarily originated from the same satellite sensor as that from which the original images used in model development came, greatly limiting the application of CT. We have developed an effective normalization method to improve the robustness of CT models when applied to images originating from different sensors and dates. A total of 965 ground-truth samples of aquatic vegetation types were obtained in 2009 and 2010 in Taihu Lake, China. Using relevant spectral indices (SI) as classifiers, we manually developed a stable CT model structure and then applied a standard CT algorithm to obtain quantitative (optimal) thresholds from 2009 ground-truth data and images from Landsat7-ETM+, HJ-1B-CCD, Landsat5-TM and ALOS-AVNIR-2 sensors. Optimal CT thresholds produced average classification accuracies of 78.1%, 84.7% and 74.0% for emergent vegetation, floating-leaf vegetation and submerged vegetation, respectively. However, the optimal CT thresholds for different sensor images differed from each other, with an average relative variation (RV) of 6.40%. We developed and evaluated three new approaches to normalizing the images. The best-performing method (Method of 0.1% index scaling) normalized the SI images using tailored percentages of extreme pixel values. Using the images normalized by Method of 0.1% index scaling, CT models for a particular sensor in which thresholds were replaced by those from the models developed for images originating from other sensors provided average classification accuracies of 76.0%, 82.8% and 68.9% for emergent vegetation, floating-leaf vegetation and submerged vegetation, respectively. Applying the CT models developed for normalized 2009 images to 2010 images resulted in high classification (78.0%–93.3%) and overall (92.0%–93.1%) accuracies. Our results suggest that Method of 0.1% index scaling provides a feasible way to apply CT models directly to images from sensors or time periods that differ from those of the images used to develop the original models.

Application of classification trees and logistic regression to determine factors responsible for lamb mortality

The aim of the presented research was to statistically analyse the survival of 20,044 Polish Merino lambs between birth and 100 day of their life, using classification trees and logistic regression. The lamb survival trait was expressed in binomial scale: 1 for survival, 0 for mortality. Two different models of the trees were developed, depending on the division criterion: they were the function of entropy and the Gini index. For comparison purposes, an additional statistical analysis was carried out using a multiple logistic regression. The quality of decision tree models and multiple regressions was compared taking into consideration the following criteria: average error function, average squared error, lift cumulative, Kolmogorov–Smirnov statistics and the area under the Receiver Operating Characteristic curve. A statistical analysis was conducted using the Enterprise Miner 6.2 software included in the SAS package. The calculated quality criteria of four models that were developed lead to the conclusion that the classification trees established based on the Gini index, and on the function of entropy, are the most accurate in defining the variability of characteristics under examination, i.e. survival of lambs up to 100 days of age. In the case of the best classification model available, i.e. a tree built using the Gini index, the ranking of variable importance, which was developed based on the “Importance” measure, leads to the conclusion that the flock, type, and the year of a lamb's birth are the most significant differentiating factors.

Application of classification trees and support vector machines to model the presence of macroinvertebrates in rivers in Vietnam

In the present study, classification trees (CTs) and support vector machines (SVMs) were used to study habitat suitability for 30 macroinvertebrate taxa in the Du river in Northern Vietnam. The presence/absence of the 30 most common macroinvertebrate taxa was modelled based on 21 physical-chemical and structural variables. The predictive performance of the CT and SVM models was assessed based on the percentage of Correctly Classified Instances (CCI) and Cohen's kappa statistics. The results of the present study demonstrated that SVMs performed better than CTs. Attribute weighing in SVMs could replace the application of genetic algorithms for input variable selection. By weighing attributes, SVMs provided quantitative correlations between environmental variables and the occurrence of macroinvertebrates and thus allowed better ecological interpretation. SVMs thus proved to have a high potential when applied for decision-making in the context of river restoration and conservation management.

Árboles de clasificación para el análisis de gráficos de control multivariantes

En control estadístico de la calidad, una de las herramientas más utilizadas son los gráficos de control. El principal problema de los gráficos de control multivariantes radica en que sólo indican que se ha producido un cambio en el proceso, pero no dice cuál o cuáles de las variables son las que originan este cambio. En la literatura especializada existen muchas aproximaciones para solucionar este problema, si bien, la más utilizada consiste en la descomposición del estadístico T2. En este trabajo se propone un método alternativo mediante la aplicación de árboles de clasificación. Los resultados obtenidos muestran que estos árboles de clasificación constituyen una buena herramienta para completar la interpretación de los gráficos de control multivariantes.

Superficial and subterranean soil erosion in Tabasco, tropical Mexico: Development of a decision tree modeling approach

Tropical countries suffer from a lack of financial resources to monitor and model sources and outcomes of soil degradation and, therefore, a scarcity of data. In this study it was tried to model superficial and subterranean water erosion based on basic environmental variables such as geological formation, soil type or vegetation cover — data which are potentially available even in developing countries. Different forms of water erosion and karst or pseudokarst formations in Tabasco, Southeastern Mexico were estimated. The study region (3500 km 2) consists of plain and hilly areas with an annual precipitation ranging from 2000 to 4000 mm. Main land use form is rangeland. The area is geologically composed of sedimentary rocks, limestone, sandstone and conglomerates. Collecting data by field observations over the entire study area a cartography of different soil loss forms such as rill and gully erosion, mass movement, sinkholes and tunnels was created including field data from available maps of geological formation, soil type, precipitation and former land use 20 years ago. Additionally, data collected in the field such as actual land use, vegetation cover and inclination were added. In the entire study area 1039 sites were affected by soil erosion with 2435 single manifestations of soil erosion. 482 sites were found with one gully each, 57 sites with erosion rills with a total of 416, 392 sites with one mass movement each, 85 sites with sinkholes with a total of 1122, and 23 sites with one tunnel each. Rendzic Leptosols over Oligocene limestone are strongly affected by sinkholes and tunnels in regions with inclinations of less than 5°. This phenomenon is explained by karstic and pseudokarstic formation. Superficial erosion forms are mainly found on rendzic Leptosols over shale sandstone and on eutric or peli-eutric Vertisols on Andesit in areas with inclinations between 5 and 30°. The application of classification trees allowed to successfully predict the occurrence of sinkholes and tunnels (degree of agreement between prediction and observation: very good to excellent). Even in areas with scarce data-bases this could allow for easy identification of high-risk areas. Predictive success of the occurrence of the different forms of superficial soil losses, however, is low. Superficial erosion seems to be mainly caused by more specific pedological factors. Nevertheless, automated induction of classification trees can be a valuable tool for preliminary data analysis and hypothesis generation in areas with lack of local expertise and can guide erosion risk mapping and soil conservation planning.

Application of classification trees in the sensitivity analysis of probabilistic model results

The complexity of some integrated-system models necessitates using a probabilistic approach to quantify uncertainty in model projections. In this work, we demonstrate how classification trees can be used to perform sensitivity analyses on probabilistic results. The classification tree technique is applied to results from the probabilistic total system performance assessment model used in the Yucca Mountain project. The technique proves effective in delineating the variables that most influence low and high outcomes.

Application of classification trees to the habitat preference of upland birds

This paper investigates the application of classification tree methodology to the bird habitat associations of the red grouse in the south Pennines of England. The effectiveness of this classification will be investigated using the kappa coefficient. Although the results for these data are disappointing, the methodology seems to have potential for this kind of problem.