Resumen en inglés

In many fields of science it is assumed that the observations under study are normally distributed, which frequently generates errors in the results since this assumption does not always coincide with the characteristics that the data actually present. Additionally, in some cases, the data can have degrees of asymmetry and/or kurtosis greater or lower than the normal distribution can capture, and in others, the data can present two or more modes. Although an alternative to deal to the particular case of the asymmetric data is reparametrization or transformation, this can generate difficulties in the interpretation of the results. In this work, new families of asymmetric distributions to fit bimodal data with positive support and high and/or low degrees of kurtosis are presented. The new families are obtained from extensions of the power-model distribution (PN) introduced by Durrans (1992), called; bimodal log-power-normal distribution and elliptical bimodal log-power-normal distribution. For the proposed models, the main properties are studied such as: the probability density function, cumulative distribution function, survival function, hazard function; moments and moment-generating function. The process of estimating the parameters is carried out by using the maximum likelihood method, and the usefulness of the proposed distributions are illustrated using a data set consisting of 85 observations on nickel concentration in soil samples that have been analyzed at the Department of Mines of the University of Atacama, Chile.