By clicking “Accept All Cookies”, you agree to the storing of cookies on your device to enhance site navigation, analyze site usage, and assist in our marketing efforts. View our Privacy Policy for more information.
leadCAST Predict helps you accelerate the classification of unknown service lines and prioritize field verifications through advanced service line material predictions powered by machine learning. leadCAST Predict is an additional feature of leadCAST LCRR compliance management solution. leadCAST Predict uses three main elements.
Data
Combine your service line inventory with parcel and census data to predict unknown materials. Our machine-learning experts use statistical techniques to prepare, clean, and improve your data to support optimal modeling.
Model
There is no one-size-fits-all modeling approach for predicting service line material. Our machine learning experts evaluate the performance and reliability of many different types of machine learning models to select the best fit for your data.
Results
Service line material predictions are viewable in an interactive dashboard. A final report describes how the machine learning model was built and tested, so that non-technical stakeholders can understand how the model works and have confidence in the model's predictions.
How much will your utility save?
Meet our data scientist team
Katie Deheer MS, MBA
(
she/her
)
Katie is the Senior Data Analyst for Trinnex and architect of leadCAST Predict, the predictive modeling module within leadCAST. Katie has over 14 years of experience in analytics-driven research, machine learning, and data visualization. She engages with utilities and regulators across the country on how to apply predictive modeling to protect public health and optimize resources.
Mark Zito, GISP, CFM
(
he/him
)
Mark is the senior product manager and architect of leadCAST and has over 15 years of experience helping utilities implement software solutions. Mark has executed over a dozen lead and copper-related projects including the award-winning Newark Lead Line Replacement Program, where he designed a data-driven solution to track the lead mitigation lifecycle.
Shervin Khazaeli, Ph.D.
(
he/him
)
Shervin is a data analytics developer and the lead data scientist for leadCAST Predict with a Ph.D. in artificial intelligence (AI) focusing on probabilistic decision-making. Shervin uses over 50 parameters in the leadCAST Predict model to optimize accuracy. He applies data science and statistical techniques to prepare, clean, and improve existing data to support optimal modeling.
Milad Omrani
(
he/him
)
Milad is a data analytics developer and data scientist for leadCAST Predict, with advanced degrees in automated manufacturing engineering and information technology engineering. He streamlines model optimization through algorithmic adjustments and hyperparameter tuning, increasing efficiency, improving accuracy, and aiding decision making.
Michaela Palmer, MS
(
she/her
)
Michaela is dedicated data professional with a M.S. in geospatial data science, passionate about leveraging her skills to analyze complex datasets and create innovative solutions to address pressing challenges in our world. With over 5 years of experience in Python, data science, and GIS, she applies her technical abilities to contribute meaningfully to projects at the intersection of science and sustainability.
The reality is that water utilities just don’t have the financial and staffing resources to verify every single service line. The challenges such as preparing for impending cybersecurity standards or finding staff, add to the pressure most utilities face every day. It can take years to complete field verifications and our communities deserve better than that. leadCAST Predict exists for one purpose only: to help you identify lead more quickly and accurately, so that you can remove and replace lines expeditiously.
Using quality data and the right modeling approach, we can successfully predict service line materials with accuracy comparable to physical verification methods.
Our data scientists work with subject matter experts to understand and explain dominant model features and optimize model performance.
An easy-to-follow predictions report
Following model configuration, our data science team prepares a report narrating how the model for your particular system was built and tested. The report provides performance metrics for understanding the accuracy and reliability of the model.
Zoomed out
Detailed view
Street view
Dynamic map and dashboard
leadCAST enables the user to choose from three different map views showing the model prediction results. These views include a zoomed-out view for a high-level view of where potential lead might exist, a detailed view showing specific neighborhoods or streets, and a street view. You’ll also have access to an exportable report that summarizes the results.
Inspection Optimizer
Evaluate whether existing field verifications are truly representative of your entire water system or if additional verifications are needed. Gain a better understanding of which properties to inspect to obtain a representative sample for unbiased and reliable predictive modeling.
By integrating Esri’s ArcGIS Lead Service Line Inventory (LSLI) solution with Trinnex’s material prediction modeling, your LCRI compliance becomes more cost-effective and streamlined. leadCAST Predict for ArcGIS allow utilities to fully leverage the suite of Esri LSLI tools and Trinnex’s predictive model using a central dataset.
Read our top predictive modeling resources:
Does Your State Accept Predictive Modeling for Inventory Development?
Confirming service line materials throughout an entire system requires field verification programs, which take time and resources to complete. However, predictive modeling focuses on leveraging information collected in the field to understand patterns for the rest of the system and predict the material of unknown services.
Predictive modeling (including machine learning) can help with inventory development through iterative, data-driven processes. But does your state accept it? Find out in this blog.
Understanding Random vs. Representative Verification Sets for Predictive Modeling
If you are a utility who is actively seeking to better understand your service line inventory, predictive modeling using a representative sample helps you make more reliable and unbiased estimates, target field verifications, and find lead service lines more efficiently. This ultimately saves time, money and leads to streamlined protection of public health.
But how do you ensure that the sample is representative? Read the blog to understand random vs. representative samples.
Journey to Clean Water: Securing Solutions to Unknowns Lead Service Lines (LSLs)
The Lead and Copper Rule Improvements (LCRI) is more than compliance — it involves improving lead sampling and corrosion control, expanding public education and customer outreach, starting lead testing in schools and childcare facilities, building a public inventory of lead service lines (LSLs), as well as replacing these LSLs, and thus addressing health risks and building trust.
In this blog, learn how utilities can make communities safer with LCRI.
The Trinnex team can walk you through how leadCAST Predict works and how it can be configured for your utility. Reach out if you have any questions or would like a full demo of leadCAST with leadCAST Predict.
Thank you! Your submission has been received!
Oops! Something went wrong while submitting the form.
.gif)
.png)
.png)