Juan José Leitón Montero

Senior data scientist with experience in hydroclimate sciences, finance, applied mathematics, statistics, and analytics. Excellent communication skills, abstract thinking, and storytelling. Fluent in R, SQL, and Python. Main interests: extreme value analysis, quantitative risk management, time series analysis, geostatistical models, causal models, and data visualization.

Curricula design for data science, data engineering, and data analysis onboarding programs (Infinite computer solutions, 2023)

Curricula for three different onboarding programs were created, based on the combination of personal experience as a lecturer, as well as workshops facilitated with different team members. The content, structure, methodology, and references from multiple courses were outlined to create a comprehensive framework to train entry-level and mid-senior associates as data analysts, data engineers and data scientists.

Customer segmentation analysis (Infinite computer solutions, 2022)

Transactional databases hosted in Snowflake were analyzed in order to produce customer profiles based on customer behavior. RFM analysis was conducted as a first approximation to the problem, producing value through key insights that led to more complex, robust segmentation models. New features were constructed, based on expert knowledge and data exploration, and clustering techniques were applied to the new datasets. Finally, customer personas were created for different verticals based on demographic information. Finally, these personas served as input for marketing campaigns and channel optimization purposes.

Uncertainty quantification for renewable energy transition (ICE, 2021)

Modeling hydroclimatic series using autoregressive models, Gaussian processes, and wavelets to quantify uncertainty in the energy portfolio of the national electricity system. Five different modeling approaches were used in order to characterize and reproduce the stochastic characteristics of a set of hydroclimatic time series. The modeled time series were evaluated in terms of the parent distribution, the distribution of extremes, and multivariate and temporal dependence structure.

National mobility study based on CDR data (ICE, 2020)

Origin-destination matrices and flow matrices were generated, based on information from CDRs, to model the effect of the sanitary measures adopted by the Ministry of Health in the fight against the COVID-19 pandemic. In addition, an algorithm was conceptualized, programmed, and implemented that allowed estimating the matrices mentioned above at the cantonal and district level. This project was a joint effort between the Ministry of Health, the Center for Research in Pure and Applied Mathematics of the University of Costa Rica, and the Costa Rican Institute of Electricity. 

Bathymetric data dashboard (ICE, 2020)

An R Shiny app was developed to store and access the bathymetric survey maps of the Costa Rican Electricity Institute's reservoirs. The dashboard allows the user to visualize and download the bathymetric maps and the elevation-area-capacity curves for more than one hundred surveys carried out by the Hydrology Department of the Costa Rican Electricity Institute during the last 18 years.

Suspended sediment rates across the Costa Rican territory (ICE, 2020)

Suspended sediment loads were estimated for more than 50 different basins across the Costa Rican territory as requested by the Central Bank of Costa Rica. First, robust regression models were fit based on +18.000 sediment samples collected by the Costa Rican Electricity Institute Hydrology Department over the past seven decades. Then, the fitted models were used to estimate sediment production rates to serve as ground truth for the calibration of the InVEST sediment model.

The project was a collaboration between the Natural Capital Project of Standford's University, the Central Bank of Costa Rica, and the Costa Rican Electricity Institute.

Time series analysis of hydropower data (ICE, 2019)

More than 15 years of 5-minutes power data from a cascade hydropower system were retrieved from the Centro Nacional de Control de Energía (CENCE) and combined into several datasets. Generalized additive models were fitted and used to estimate the energy production of the downstream plant as a function of both the river discharge and the energy production of the upstream plant.

The models were used to approximate the hypothetical energy production of the downstream plant before its construction. The main objective was to assess if the current energy production values were due to a change in the operation regime of the upstream plant or due to climate variability.

Statistical analysis of multi-model climate projections with a Bayesian hierarchical model over Europe (ETHZ, 2017)

A hierarchical Bayesian model was used to analyze the seasonal temperature and precipitation projections over the PRUDENCE regions of the CH2018 multi-model ensemble (RCP8.5). The implementation of this model expands the work done by Kerkhoff (2014), Tay (2016), and Künsch (2017) by evaluating both temperature and precipitation variables for every region-season combination.

Posterior distributions for the parameters associated with bias assumption coefficients, climatological means, inter-annual variability, and additive bias were estimated. Similarly, for five different time horizons, climate change estimates were calculated with respect to 1995. A generalized variation pattern was found for temperature along all the region-season combinations analyzed, while season-dependent and region-dependent patterns were identified for precipitation.

The absolute additive bias reduction due to dynamical scaling was evaluated by comparing the bias components associated with the RCM-GCM chains and their corresponding drivers. Results were assessed in terms of the probabilities of reducing at least 20% in the said component, and region-season-chain combinations were classified based on this value.

• Leitón-Montero, J.J. (2017). Statistical analysis of multi-model climate projections with a Bayesian hierarchical model over Europe. Master thesis. ETH Zürich

Report on hydrology and sediments for the Costa Rican river basins draining directly to the San Juan River (ICE, 2013-2015)

Conceptualizing, planning, and conducting hydrological and sedimentological studies in the San Juan river basin to determine the average liquid and solid discharge regimes. Building a sediment-budged for the primary basin system using the Universal Soil Loss Equation (USLE). Analyzing +60.000 storms, 2350 sediment samples, and +60 years of hydrological data.

• Gómez-Delgado, F., Leitón-Montero, J.J & Aguilar-Cabrera, C.A. (2013). Report on hydrology and sediments for the Costa Rican river basins draining directly to the San Juan River. San José, Costa Rica: Costa Rican Electricity Institute. In Government of Costa Rica (Ed.), Dispute concerning the construction of a road in Costa Rica along the San Juan River (Nicaragua v. Costa Rica) Counter-Memorial of Costa Rica Vol II (pp. 133-290). San José, Costa Rica.
• Leitón-Montero, J.J. & Jiménez-González, D. (2014). Second report on hydrology and sediments for the Costa Rican river basins draining directly to the San Juan River. San José, Costa Rica: Costa Rican Electricity Institute. In Government of Costa Rica (Ed.), Dispute concerning the construction of a road in Costa Rica along the San Juan River (Nicaragua v. Costa Rica) Rejoinder of Costa Rica Vol III (pp. 45-126). San José, Costa Rica.
• Technical counseling to the Costa Rican legal team during the oral proceedings before the International Court of Justice at The Hague, The Netherlands (April 2015).

Methodology for the determination of adaptive flow (ICE, 2014)

Regionalization of hydrological descriptors (mean annual runoff, variance, and flow duration curves) in three Costa Rican basins using the hydrostochastic interpolation approach.

• Krasovskaia, I., Gottschalk, L., Leitón-Montero, J.J., Zuñiga-Mora, J.A. & Leblois, E. (2014). REGINA - Regionalisation of hydrological descriptors. In I. Krasovskaia & L. Gottschalk (Eds.), Compensatory Flow Phase 2: Methodology for the determination of adaptive flow - Final Report. (English edition, pp. 25-79). San José, Costa Rica: Costa Rican Electricity Institute.

Change in the runoff pattern in the upper basin of the Tempisque River under climate change scenarios (UCR, 2011)

Statistical downscaling of precipitation and temperature monthly time series from General Circulation Model's (GCM) was conducted via Principal Component  Analysis (PCA), Canonical Correlation Analysis (CCA), and Delta Scheme. The downscaled time series served as input to a water balance model, and the change in the mean value of runoff was estimated for two different time horizons.

• Leitón-Montero, J.J (2011). Change in the runoff pattern in the upper basin of the Tempisque River under climate change scenarios (in Spanish). Licentiate thesis. The University of Costa Rica.