Count Regression Models Research Articles

Count Data Regression Modelling: An Application to Monkeypox Confirmed Cases

Introduction: With the presence of COVID 19, some countries also faced an increase in number of cases due to Monkeypox virus. The main aim of this research was to investigate whether it is possible to fit count data regression models to predict the daily incidence of Monkeypox confirmed cases.
 Methods: In this study we have used two types of traditional count regression models like Poisson regression model and Negative binomial regression model using identity and logarithmic link function. Since our data was overdispersed, Negative binomial regression model with logarithmic link function fitted well as compared to other models. The parameters were estimated using SPSS, version 23.0.
 Results: The Negative Binomial Regression model with logarithm function fits well to the data related to Monkeypox cases. Therefore, the model shows that majority of the countries like Brazil, Canada, France, Germany, Peru, Spain, United Kingdom and United States of America shows significant decrease in number of cases with respect to time. The prediction line was plotted using this model where the line predicts well about the daily Monkeypox cases reported by different countries.
 Conclusion: From our study, we concluded that the count data regression model can be used widely to predict the incidence of any disease. The countries like Canada and Brazil have largest and smallest slope coefficient which shows maximum and minimum decrease in expected number of cases confirmed each day respectively.

Bernoulli Poisson Moment Exponential Distribution: Mathematical Properties, Regression Model, and Applications

We introduce a new flexible count distribution by combining Bernoulli and Poisson moment exponential (PMEx) distributions. The new model named the Bernoulli PMEx distribution. Some mathematical properties are studied, including the hazard rate function, moments, moment generating function, probability generating function, and dispersion index. A count regression model is also proposed based on this distribution. The maximum likelihood estimation method is used to estimate the model parameters. In the end, three datasets from different fields are utilized for application purposes. The findings show that the new model efficiently analyzed these datasets as compared to Poisson, discrete Pareto, discrete Rayleigh, discrete Burr‐Hatke, and discrete inverted Topp–Leone distributions.

Zero-Inflated Count Regression Models in Solving Challenges Posed by Outlier-Prone Data; an Application to Length of Hospital Stay.

Ignoring outliers in data may lead to misleading results. Length of stay (LOS) is often considered a count variable with a high frequency of outliers. This study exemplifies the potential of robust methodologies in enhancing the accuracy and reliability of analyses conducted on skewed and outlier-prone count data of LOS. The application of Zero-Inflated Poisson (ZIP) and robust Zero-Inflated Poisson (RZIP) models in solving challenges posed by outlier LOS data were evaluated. The ZIP model incorporates two components, tackling excess zeros with a zero-inflation component and modeling positive counts with a Poisson component. The RZIP model introduces the Robust Expectation-Solution (RES) algorithm to enhance parameter estimation and address the impact of outliers on the model's performance. Data from 254 intensive care unit patients were analyzed (62.2% male). Patients aged 65 or older accounted for 58.3% of the sample. Notably, 38.6% of patients exhibited zero LOS. The overall mean LOS was 5.89 (± 9.81) days, and 9.45% of cases displayed outliers. Our analysis using the RZIP model revealed significant predictors of LOS, including age, underlying comorbidities (p<0.001), and insurance status (p=0.013). Model comparison demonstrated the RZIP model's superiority over ZIP, as evidenced by lower Akaike information criteria (AIC) and Bayesians information criteria (BIC) values. The application of the RZIP model allowed us to uncover meaningful insights into the factors influencing LOS, paving the way for more informed decision-making in hospital management.

Predictors of frequency of 1-year readmission in adult patients with diabetes

Diabetes mellitus (DM) is the third most common chronic condition associated with frequent hospital readmissions. Predictors of the number of readmissions within 1 year among patients with DM are less often studied compared with those of 30-day readmission. This study aims to identify predictors of number of readmissions within 1 year amongst adult patients with DM and compare different count regression models with respect to model fit. Data from 2008 to 2015 were extracted from the electronic medical records of the National University Hospital, Singapore. Inpatients aged ≥ 18 years at the time of index admission with a hospital stay > 24 h and survived until discharge were included. The zero-inflated negative binomial (ZINB) model was fitted and compared with three other count models (Poisson, zero-inflated Poisson and negative binomial) in terms of predicted probabilities, misclassification proportions and model fit. Adjusted for other variables in the model, the expected number of readmissions was 1.42 (95% confidence interval [CI] 1.07 to 1.90) for peripheral vascular disease, 1.60 (95% CI 1.34 to 1.92) for renal disease and 2.37 (95% CI 1.67 to 3.35) for Singapore residency. Number of emergency visits, number of drugs and age were other significant predictors, with length of stay fitted as a zero-inflated component. Model comparisons suggested that ZINB provides better prediction than the other three count models. The ZINB model identified five patient characteristics and two comorbidities associated with number of readmissions. It outperformed other count regression models but should be validated before clinical adoption.

On the need to address fixed-parameter issues before applying random parameters: A simulation-based study

Count regression models have been applied to model expected crash frequency at individual roadway locations. Random parameters have been increasingly integrated into these models to account for unobserved heterogeneity. However, the introduction of random parameters might also mask issues in the model specification, leading to inaccurate relationships and model interpretation. Two of these specification-related issues are: (1) not considering the appropriate functional form of explanatory variables; and, (2) ignoring the best set of significant explanatory variables. To better examine the need for careful model specification, this study uses synthetic data to demonstrate that the consideration of random parameters does not address the two model specification issues identified. The results from the simulation study illustrate that (a) model specification issues cannot be circumvented by random parameters alone and (b) random parameter models including the exhaustive set of explanatory variables available offer significant model improvements.

Application of a count regression model to identify the risk factors of under-five child morbidity in Bangladesh.

Bangladesh has seen a significant decline in child mortality in recent decades, but morbidity among children <5y of age remains high. The aim of this analysis was to examine trends and identify risk factors related to child morbidity in Bangladesh. This analysis is based on data from four successive cross-sectional Bangladesh Demographic and Health Surveys for the years 2007, 2011, 2014 and 2017-18. Several count regression models were fitted and the best model was used to identify risk factors associated with morbidity in children <5y of age. According to the results of the trend analysis, the prevalence of non-symptomatic children increased and the prevalence of fever, diarrhoea and acute respiratory infections (ARIs) decreased over the years. The Vuong's non-nested test indicated that Poisson regression could be used as the best model. From the results of the Poisson regression model, child age, sex, underweight, wasted, stunting, maternal education, wealth status, religion and region were the important determinants associated with the risk of child morbidity. The risk was considerably higher among women with a primary education compared with women with a secondary or greater education in Bangladesh. This analysis concluded that child morbidity is still a major public health problem for Bangladesh. Thus it is important to take the necessary measures to reduce child morbidity (particularly fever, diarrhoea and ARI) by improving significant influencing factors.

A new discrete distribution with applications to radiation, smoking, and health data

In this study, a flexible discrete distribution is introduced by compounding Epanechnikov-exponential and Poisson distributions. Many mathematical features of the distribution are obtained. Several point estimators of the unknown parameter are proposed. The Monte Carlo simulation study is carried out to compare these estimators. Three methods are proposed to construct the approximate confidence intervals for the distribution parameter. A novel zero-inflated regression model is given based on the proposed discrete distribution as an alternative to zero-inflated geometric and zero-inflated Poisson models. The usability of the distribution and the count regression model is demonstrated by some practical data analyses, such as the number of dicentric chromosomes observed after exposure to radiation at doses of 0.600, survey data on health, and smoking habits.

Zero-inflated Poisson regression analysis of factors associated with under-five mortality in Ethiopia using 2019 Ethiopian mini demographic and health survey data

Background Remarkable reduction in global under-five mortality has been seen over the past two decades. However, Ethiopia is among the five countries which account for about half (49%) of all under-five mortality worldwide. This study aimed at identifying factors associated with under-five children mortality in Ethiopia using the 2019 Ethiopia mini demography and health survey data. Methods The most recent national representative demography and health survey data of Ethiopia, 2019 Ethiopia mini demography and health survey data, were used for this study. Count data regression models were applied to identify the factors associated with under-five children mortality. Statistical significance was declared at P-value less than 0.05. Results Zero-Inflated Poisson (ZIP) regression model was found to be the best model compared to other count regression models based on models comparison Criteria. The ZIP model revealed that decreased risk of under-five mortality was associated with mothers aged 25–34 years, unmarried mothers, mothers delivered in health facility, mothers used Pill/IUD, mothers who had larger number of children at home whereas increased risk of under-five mortality was associated with older mothers at their first births, mothers from rural areas, mothers travel for 1–30 min and &gt;30 min to get drinking water, mothers used charcoal and wood, children with higher birth order and multiple births. Conclusions In this study, place of residence, region, place of delivery, religion, age of mother, mother’s age at first birth, marital status, birth order, birth type, current contraceptive type used, type of cooking fuel, time to get drinking water, and number of children at home were statistically significant factors associated with under-five mortality in Ethiopia. Thus, the Ethiopian Ministry of Health and other concerned bodies are recommended to encourage mothers to deliver at health institutions, give awareness for mothers to use Pill/IUD contraceptive type, and facilitate rural areas to have electricity and drinking water near to homes so as to minimize the under-five mortality to achieve the sustainable development goal.

Factors determining antenatal care utilization among mothers of deceased perinates in Ethiopia.

Receiving adequate antenatal care (ANC) had an integral role in improving maternal and child health outcomes. However, several factors influence the utilization of ANC from the individual level up to the community level factors. Thus, this study aims to investigate factors that determine ANC service utilization among mothers of deceased perinate using the proper count regression model. Secondary data analysis was performed on perinatal death surveillance data. A total of 3,814 mothers of deceased perinates were included in this study. Hurdle Poisson regression with a random intercept at both count-and zero-part (MHPR.ERE) model was selected as a best-fitted model. The result of the model was presented in two ways, the first part of the count segment of the model was presented using the incidence rate ratio (IRR), while the zero parts of the model utilized the adjusted odds ratio (AOR). This study revealed that 33.0% of mothers of deceased perinates had four ANC visits. Being in advanced maternal age [IRR = 1.03; 95CI: (1.01-1.09)], attending primary level education [IRR = 1.08; 95 CI: (1.02-1.15)], having an advanced education (secondary and above) [IRR = 1.14; 95 CI: (1.07-1.21)] and being resident of a city administration [IRR = 1.17; 95 CI: (1.05-1.31)] were associated with a significantly higher frequency of ANC visits. On the other hand, women with secondary and above education [AOR = 0.37; 95CI: (0.26-0.53)] and women who live in urban areas [AOR = 0.42; 95 CI: (0.33-0.54)] were less likely to have unbooked ANC visit, while women who resided in pastoralist regions [AOR = 2.63; 95 CI: (1.02-6.81)] were more likely to have no ANC visit. The uptake of ANC service among mothers having a deceased perinate was determined by both individual (maternal age and educational status) and community (residence and type of region) level factors. Thus, a concerted effort is needed to improve community awareness through various means of communication by targeting younger women. Furthermore, efforts should be intensified to narrow down inequalities observed in ANC service provision due to the residence of the mothers by availing necessary personnel and improving the accessibility of service in rural areas.

Count regression models analysis of factors affecting antenatal care utilization in Ethiopia: original article.

Antenatal care (ANC) reduces maternal and newborn mortalities and makes it easier to find infections early and prevent them from progressing. This study aimed to identify factors affecting ANC utilization in Ethiopia via the count regression model. The data for the study was drawn from the 2019 Ethiopian Mini Demographic and Health Survey dataset. Count regression models, such as Poisson, Negative Binomial (NB), Zero-Inflated Poisson (ZIP), and Zero-Inflated Negative Binomial (ZINB) models, were used to determine the factors influencing ANC utilization. A total of 3962 women were included in the study. Only about 42% of women used the WHO-recommended number of ANC of a minimum of four visits. The ZIP model was outperforming to fit the data as compared to other count regression models. Rural residence (IRR=0.8832, 95% CI: 0.8264-0.9440), being resident of the Somalia region (IRR=0.4762, 95% CI: 0.3919-0.5785), SNNPR (IRR=0.8716, 95% CI: 0.7913-0.9600), and Gambela Region (IRR=0.7830, 95% CI: 0.7063-0.8680), being Muslim (IRR=0.9384, 95% CI: 0.8876-0.9921) decrease the ANC utilization. Contrarily, Addis Ababa residence (IRR=1.1171, 95% CI: 1.0181-1.2259), primary education (IRR=1.1278, 95% CI: 1.0728-1.1855), secondary and higher education (IRR=1.2357, 95% CI: 1.1550-1.3220), middle wealth index (IRR=1.0855, 95% CI: 1.0095-1.1671) and rich wealth index (IRR=1.0941, 95% CI: 1.0152-1.1790) increase the ANC utilization. The ZIP model best fitted the data compared to others. The study revealed that being poor, rural resident, uneducated, Somalia region resident, SNNPR resident, Gambela region resident, and Muslim were factors associated with lower ANC utilization. Thus, health education is needed to be given for mothers with no education. In addition, building a strong awareness-creation program regarding ANC is required for rural residents so as to improve the level of ANC utilization in Ethiopia.

FOOD SUPPLEMENTATION INCREASES MULE DEER (Odocoileus hemionus eremicus) RETENTION AT THE RANCHO SAN HUBERTO WILDLIFE CONSERVATION MANAGEMENT UNIT, SONORA, MEXICO

Artificial food and water supplementation programs are common in Wildlife Conservation Management Units (UMAs) in arid northern Mexico to attract and retain mule deer as quality trophies. However, implementing these actions represents a high economic cost, and there is no convincing evidence of their effectiveness. This study evaluated the effect of food and water supplementation by capture-mark-recapture using camera traps at the UMA San Huberto, Sonora, Mexico. During the winter of 2015-2016, male mule deer counts and detection histories were collected at 62 photo-trapping stations spread across four supplementation treatments: 1) water and food, 2) water, 3) food, and 4) no supplementation. In addition, to assess the impact of natural vegetation on trough and feeder utilization on the ranch, the Normalized Difference Vegetation Index was calculated using Landsat 8 satellite images. A log-Poisson regression model for deer counts per photo-trapping station found that food supplementation increased mule deer numbers by 42 % when compared to stations without food supplementation. Vegetation biomass had no effect on the frequency of use of feeders and waterers in any of the treatments. According to the multistate model for detection histories, male mule deer at waterers receiving food-only supplementation are more likely to remain in that treatment than to move to either of the other two treatments (water and feed or water). This finding suggests that food-only supplementation is more effective in retaining male mule deer.

Factors associated with the number of months of delaying in multiple sclerosis diagnosis: Comparison of count regression models.

Background: It may take a long time to diagnose multiple sclerosis (MS) since the emergence of primary symptoms. This study aimed to use count regression models to compare their fit and to identify factors affecting delay in the diagnosis of MS. Methods: Data were collected from the Nationwide MS Registry of Iran (NMSRI) for Mazandaran Province, Iran, using census sampling until April 2022. The four models of Poisson regression, negative binomial (NB) regression, zero-inflated Poisson (ZIP) regression, and zero-inflated negative binomial (ZINB) regression were used in this study. Results: In this study on 2894 patients, 74.0% were women, and 8.5% had a family history of MS. The mean ± standard deviation (SD) of the patients' age was 34.96 ± 9.41 years, and the mean delay in diagnosis was 12.32 ± 33.26 months, with a median of 0 (Q1-Q3: 0-9). The NB regression model showed the best performance, and factors, including a history of hospitalization and the year of symptom onset, had significant effects on a delayed diagnosis. Besides, the Expanded Disability Status Scale (EDSS) score was significantly different before and after 2017; it was also associated with sex, type of MS, and history of hospitalization. Conclusion: The mean diagnostic delay and the mean age of MS diagnosis are critical in Mazandaran Province. Patients with MS develop the disease at an early age and are diagnosed with a long delay. The time of symptom onset is a significant factor in the diagnosis of MS, and in recent years, there have been improvements in the diagnostic process.

Determinants of sexually transmitted infections among female sex workers in Ethiopia: a count regression model approach.

Sexually transmitted infections (STIs) remain a major public health problem worldwide, with the burden of these infections being high among female sex workers (FSWs), who are often not aware of their infection status. This study aimed to determine the factors that are associated with the number of STIs among FSWs in Ethiopia. A cross-sectional bio-behavioral study involving respondent-driven sampling (RDS) was conducted among 6,085 FSWs in 16 towns in Ethiopia. The hurdle Poisson regression model was fitted using STATA Version 16.2. The incident rate ratio and adjusted odds ratio with a 95% confidence interval were employed to show the strength and direction of the association. A p-value of ≤0.05 was used as a threshold for statistical significance. At least one STI was identified in 1,444 (23.64%) of the FSWs. Age group 35-49 years [IRR = 2.32; 95% CI (1.43, 3.74)], forced first sex [IRR = 1.32; 95% CI (1.01, 1.74)], condom breakage [IRR = 1.32; 95% CI (1.01, 1.74)], and a history of depression [IRR = 1.55; 95% CI (1.12, 2.18)] increase the number of STIs. FSWs aged 25-34 years [AOR = 2.99; % CI (2.54, 3.52)] and 35 = 59 years [AOR = 8.05; % CI (6.54, 9.91)], who were selling sex for 5-10 years [AOR = 1.30; 95% CI (1.1, 1.55)], and above 11 years [AOR = 1.21; 95% CI (1.03, 1.43)] were more likely to get STIs. STIs are common in Ethiopia. The covariates age, educational status, monthly income, condom failure, age at the first sexual encounter, and long duration of sexual practice are significant predictors of STIs. Health interventions among FSWs need to include awareness generation about the prevention and control of STIs and address the determinants identified in this analysis.

A New Two-Parameter Discrete Distribution for Overdispersed and Asymmetric Data: Its Properties, Estimation, Regression Model, and Applications

A novel discrete Poisson mixing probability distribution with two parameters has been developed by combining the Poisson distribution with the transmuted moment exponential distribution. It is possible to deduce several mathematical properties, such as the moment-generating function, ordinary moments, moments about the mean, skewness, kurtosis, and the dispersion index. The maximum likelihood estimation method is utilized to estimate the model’s parameters. A thorough simulation study is utilized to determine the behavior of the generated estimators. Estimating model parameters using a Bayesian methodology is another primary topic of this research. The behavior of Bayesian estimates is evaluated by first charting the trace, then generating 1,005,000 iterations of the Markov chain Monte Carlo method. In addition to this, we suggest a new count regression model that uses Poisson and negative binomial models in an alternating fashion. In conclusion, asymmetric datasets derived from various research areas are utilized for practical applications.

Intensity of adaptations to heat stress in poultry farms: A behavioural analysis of farmers in Ondo state, Nigeria

The detrimental consequences of heat stress due to high ambient temperatures, particularly in the poultry industry, have led to the invention of several adaptation strategies. However, there is still limited information on the intensity of adaptations and the likely behavioural factors that influence farmers' decisions. Thus, understanding the practical adaptation behaviours of poultry farmers would improve our knowledge of the fundamental mechanisms for developing effective interventions. To fill this void, using a count data model, the study empirically examines the farmers' behavioural factors and the intensity of heat stress adaptation strategies’ adoption among poultry farmers in Ondo State, Nigeria. The data were drawn from a survey of 150 poultry farmers using a multistage sampling procedure. The empirical results show that the majority of the farmers perceived an increase in temperature, frequently experienced heat stress, and believed that heat stress is induced by climate change. An average of six adaptation strategies were simultaneously adopted to mitigate heat stress in the area. The results of the count regression model reveal that farm-level factors such as permanent water sources, the quantity of feed, and bird stock density exert a significant effect on the intensity of adaptations. Climate-related factors such as access to climate information, training participation, perceived increases in temperature, attitudes toward climate change, and motives for adoption have a significant behavioural effect on the intensity of adaptations. Likewise, variables such as poultry farming experience, educational status, and access to credit are accounted for as socioeconomic behavioural factors that influence the intensity of adopting heat stress adaptation strategies in the area. This concludes that behavioural factors are crucial in addressing heat stress adaptations and assisting in improving environmental management, which would form a key variable in the policy interventions.

EM estimation for bivariate mixed poisson INAR(1) claim count regression models with correlated random effects

This article considers bivariate mixed Poisson INAR(1) regression models with correlated random effects for modelling correlations of different signs and magnitude among time series of different types of claim counts. This is the first time that the proposed family of INAR(1) models is used in a statistical or actuarial context. For expository purposes, the bivariate mixed Poisson INAR(1) claim count regression models with correlated Lognormal and Gamma random effects paired via a Gaussian copula are presented as competitive alternatives to the classical bivariate Negative Binomial INAR(1) claim count regression model which only allows for positive dependence between the time series of claim count responses. Our main achievement is that we develop novel alternative Expectation-Maximization type algorithms for maximum likelihood estimation of the parameters of the models which are demonstrated to perform satisfactory when the models are fitted to Local Government Property Insurance Fund data from the state of Wisconsin.

Does police patrol in large areas prevent crime? Revisiting the Kansas City Preventive Patrol Experiment

AbstractResearch SummaryThe Kansas City Preventive Patrol Experiment (KCPPE) was seen by its developers to have produced “consistent evidence of the lack of effects of any consequence on crime,” a conclusion that was to have a strong impact on assumptions about police patrol for almost half a century. We identified the original official crime data from the KCPPE, and reanalyzed outcomes focusing on a comparison of the “proactive” versus “control” beats (“reactive beats” were criticized because of violations of treatment integrity); examining broad categories of crime (to increase statistical power); and using count regression models. Our findings are not unequivocal, but point to modest impacts of police patrol on crime in police beats.Policy ImplicationsOur findings suggest that lessons drawn for half a century from the KCPPE need to be revisited. The KCPPE does not show that police patrol in large areas has no influence on crime, and this finding is consistent with several more recent studies. At the same time, we note that the effects of patrol in the KCPPE using our analysis strategy, and those found in other studies of preventive patrol in larger areas, are about half that found in hot spots policing studies. This suggests that police agencies ideally should invest in focused hot spots policing initiatives. However, absent an ability to manage such initiatives, or the crime analysis capabilities to identify crime hot spots routinely, simpler preventive patrol schemes to utilize uncommitted patrol time can be seen as potentially effective in preventing crime.

Identifying predictors of tooth loss using a rule-based machine learning approach: A retrospective study at university-setting clinics.

This study aimed to identify predictors associated with tooth loss in a large periodontitis patient cohort in the university setting using the machine learning approach. Information on periodontitis patients and 18 factors identified at the initial visit was extracted from electronic health records. A two-step machine learning pipeline was proposed to develop the tooth loss prediction model. The primary outcome is tooth loss count. The prediction model was built on significant factors (single or combination) selected by the RuleFit algorithm, and these factors were further adopted by the count regression model. Model performance was evaluated by root-mean-squared error (RMSE). Associations between predictors and tooth loss were also assessed by a classical statistical approach to validate the performance of the machine learning model. In total, 7840 patients were included. The machine learning model predicting tooth loss count achieved RMSE of 2.71. Age, smoking, frequency of brushing, frequency of flossing, periodontal diagnosis, bleeding on probing percentage, number of missing teeth at baseline, and tooth mobility were associated with tooth loss in both machine learning and classical statistical models. The two-step machine learning pipeline is feasible to predict tooth loss in periodontitis patients. Compared to classical statistical methods, this rule-based machine learning approach improves model explainability. However, the model's generalizability needs to be further validated by external datasets.

Factors Associated with Women Fertility in Bangladesh: Application on Count Regression Models

Background: The current total fertility rate in Bangladesh is now 2.3 births per woman, which is still above the replacement level of 2.1. Objective: The main objective of this study was to identify potential factors associated with fertility transition in Bangladesh. Methods: This study applied several regression models to find the best-fitted model to determine factors associated with the number of children ever-born in Bangladesh and utilize data from the 2019 Bangladesh Multiple Indicator Cluster Survey. Results: Based on the principles of the AIC, BIC, and Vuong tests, the best-fit model was the Hurdle- Poisson regression model compared to other models. Findings based on the Hurdle Poisson regression result revealed that the number of children increases with the increase of women’s age, but the number of children declines if the education status of women as well as their delayed marriage increases. Women who had secondary or higher education were less likely to have children than illiterate women. Similarly, division, residential area, wealth index, women’s functional difficulties, prenatal care, and migration have significantly influenced the number of children ever born. Conclusion: Based on the findings, the study suggests that fertility can be decreased by improving female education, minimizing early marriage, and eliminating poverty for all ever-married women who were particularly live in rural areas of the Chattogram and Sylhet divisions in Bangladesh. Such steps would be the largest contribution to a future reduction in fertility rates in Bangladesh.

Count Regression and Machine Learning Techniques for Zero-Inflated Overdispersed Count Data: Application to Ecological Data

The aim of this study is to investigate the overdispersion problem that is rampant in ecological count data. In order to explore this problem, we consider the most commonly used count regression models: the Poisson, the negative binomial, the zero-inflated Poisson and the zero-inflated negative binomial models. The performance of these count regression models is compared with the four proposed machine learning (ML) regression techniques: random forests, support vector machines, k-\\documentclass[12pt]{minimal} \\usepackage{amsmath} \\usepackage{wasysym} \\usepackage{amsfonts} \\usepackage{amssymb} \\usepackage{amsbsy} \\usepackage{mathrsfs} \\usepackage{upgreek} \\setlength{\\oddsidemargin}{-69pt} \\begin{document}$$k-$$\\end{document}nearest neighbors and artificial neural networks. The mean absolute error was used to compare the performance of count regression models and ML regression models. The results suggest that ML regression models perform better compared to count regression models. The performance shown by ML regression techniques is a motivation for further research in improving methods and applications in ecological studies.