Blog

Latest News and Updates

4 tips for establishing an automated risk management process 

When your organization transitions to machine learning credit underwriting technologies, it is very beneficial to use an automated risk management process. It will help your organization keep up with changes in the market, being able to explain your business to regulators and make your organization more efficient. So what should you think about when going for an automated risk management process?

1: Document the data

It all begins with data. It is important to track, analyse and measure all the data that has been used to create the model. Understanding why each feature is included in the model and how missing data is imputed. We also recommend to document every model you have tried during model development, along with its performance and results.

2: Understand model – technically

Before deploying your machine learning credit underwriting model into production, you must understand how your model operates. This includes understanding the global variable/feature importance and how the variables interact with each other. Is your model trained on data which includes specific policy rules? Then you have to know your model’s blindspot and weaknesses. At last what is your performance metric for scalability, how many loan application can you model handle in a live environment?

3: Understand model – businesswise

The economic impact of the model must be quantifiable too, and questions such as:
How does the model impact revenue?
Will the model change approval rate or default rate?
How does the model affect the distribution of scores?  
Without having these answers, the model you put into production can behave very differently compared to the test results you have.

4 Monitor your model

A robust risk management process, will help you minimise the number of surprises a model put into production can have. However it cannot completely eliminate the unknown. Therefore it must exist a process where you can quickly validate and follow the model and have the possibility to quickly make changes if something goes wrong. At top of that, monitoring the model’s variable distribution over time can help you alert market changes, foresee the next macroeconomic shock and have a competitive advantage during rough environment

 

An Automated risk management process is a must for financial institutions when  utilizing the benefits of artificial intelligence in credit underwriting. Following these best practices will help you ensure that your model is powerful, accurate and robust. Allowing you to to identify better borrowers, which artificial intelligence in credit underwriting brings.

Are you ready to use artificial intelligence in combination with a robust risk management process?

Contact partnership@evispot.ai for more information

read more
Tomas Sellden4 tips for establishing an automated risk management process 

Evispot’s AI-platform has been further developed to ease the SME credit assessment

It is well-known that a lot of SMEs (small-medium enterprises) struggles to find financial help. To apply for a loan, a lot of administration work is required and the final result is either a rejected loan or a loan offer with huge interest rate. Meanwhile, it is a huge market, only in Sweden it exists about 1,5 million companies and 99% of them are less than 50 employees.

So how come such a big market has issues finding capital?

One reason is that it is very difficult to understand if a company is well-being or not. There are many factors to consider and a lot of data to go through and analyse. Since there are so many perspectives to consider, credit assessing companies has always required a lot of administration and manual work. The result is that many SMEs are having a hard time finding suitable financial help and the few companies that are granted loans have to pay huge interest rate, since the cost for manually assessing the loan application is high.

With access to more data and technology to analyse and understand data, there exist tools which can ease  the credit assessment for the creditors.

Do you want to know how the Evispot AI-platform can help you ease the credit assessment of company credits? Contact: partnership@evispot.ai

The Evispot AI-platform is based on a complex array of scoring algorithms developed on hundreds of thousands of data points from different sources such as accounting data, annual reports, industry specific trends and can include your bank´s historical data to ease the job for a creditor to understand if a company is good or not.

read more
Tomas SelldenEvispot’s AI-platform has been further developed to ease the SME credit assessment

Evispot helps Ativo take smarter decisions and decrease the number of manual decisions

In the beginning of 2018 Ativo Finans integrated their IT-system to Evispot’s credit decision platform. Since integration Ativo Finans has used the platform to take smarter credit decisions and decrease the number of manual decisions.

 

 

We would definitely recommend Evispot. From our very first interaction to onboarding, it’s always been a great experience. It’s an excellent credit decision platform and I’m glad we chose Evispot.” 

– Erik Finné, CCO/CIO at Ativo Finans

 

About Ativo Finans

Ativo Finans is a Swedish finance and credit company providing factoring services and loans to Swedish small and medium sized companies. They deliver high customer value through digitized, automated operations combined with personalized customer support. Ativo Finans also provides factoring services under the brand Factoringgruppen.

About Evispot

Evispot is a credit decision platform that helps you as a creditor to truly understand your customers. With the power of artificial intelligence, the platform reveals the complete picture of your customer behavior, enabling you to do more accurate credit decisions.

read more
Tomas SelldenEvispot helps Ativo take smarter decisions and decrease the number of manual decisions

How Artificial Intelligence (AI) Improves Credit Underwriting

How can you as a creditor increase your lending portfolio without adding risk?

The answer is simple: Find the good borrowers which traditional credit scoring techniques oversees and rejects. And they do exist –  traditional methods, such as logistic regression, has a hard time accurately classifying some populations, such as those with little credit history or people who have had past credit issues.

Traditional credit agencies usually categorize a applicant into a risk group between 1-10, which is a good indicator for many of the applicants. However each risk category will have some applicants who are misclassified –  which should be classified in a higher or lower risk class. In sense, just because a person has a low credit score from a credit agencies it does not necessarily mean that he or she is a bad payer. And it goes goes to other way around as well, high scores doesn’t  necessarily mean you are a good payer.

By utilizing the power of AI we can get a complete picture of your applicants. Using more data, and more importantly, by understanding and capturing the interactions between the data is it possible to understand who will pay back and not. Resulting in more accurate credit decisions for you.

As can be seen in the picture above, an AI model will swap some of the borrowers to a higher risk class and some to the lower risk class. The end result is a credit decision model which helps you to either more approvals without increasing risk, or less risk with the same approval rate.

Evispot’s AI-platform help financiers to make use of more profitable AI credit models. If you are ready to take the next step contact us at partnership@evispot.ai

read more
Tomas SelldenHow Artificial Intelligence (AI) Improves Credit Underwriting
Traits1-1.svg

Evispot Traits – Transparent and Explainable Artificial Intelligence

It is no secret that Artificial Intelligence (AI) improves credit underwriting. It allows creditors taking advantage of the vastly more data and switching from point-in-time solutions to change-over-time solutions – resulting in credit models with higher accuracy than ever before. With a traditional, point-in-time, model is the applicant’s number of historical payment remarks calculated and measured. With an AI, change-over-time, model is it possible to understand the trends and spot the differences between applicants who have gained payment remarks recently compared to the ones who got them several years ago. Hence, AI allows creditors to identify the difference between the applicants who are on an upgoing and downgoing trend in their creditworthiness.  Applicants who otherwise would have been seen as the same. This is only one example of how AI allows creditors to identify risk more accurately.  

 

So, why hasn’t AI been highly adopted in credit underwriting?
Due to a regulated industry and since AI-models has historically been black-box solutions.

 

 

Demystifying the Black Box

We at Evispot have decided to transform credit underwriting by using the power of AI combined with transparency and explainability. Our mission is to deliver solutions which elevate creditors to new heights – our customers should be able to meet the demands of transparency while still having the accuracy and performance of AI.

At Evispot we have demystified the black-box, through three steps: transparency, explainability and provability.

Figure 1: Breaking down the black box in three steps. 

Transparency, explainability and probability are our three technical key indicators helping us and our customers understand the full spectrum of how the model reasons, predicts, and – simply put – works. Transparency is about the AI-models high-level performance, the holistic view of its characteristics to help the creditors understand both the impact of the AI-model and the reasoning behind the model. Explainability is about helping the creditor understand each decision separately. Lastly, provability is used as a quality measurement of each prediction, to ensure the highest possible quality in each credit decision.

 

 

Introducing Evispot Traits

The solution we have developed is called Evispot Traits. When Evispot Traits sees an AI-model it returns a transparent and explainable description of the model. An AI-model is trained on a set amount of historical loans, good and bad ones with a set number of variables for each loan. The model has learnt itself which variables and combinations of variables that are significant for good and bad payers. Evispot Traits calculates how much a specific variable contributed to an actual historical event.  This allows us to create a score of each variable’s contribution to each historical loan.

This number is called Evispot Score, it varies between -1 to 1 and each historical application score is plotted as a dot (see figure 2 below). The horizontal position of the dot is the impact of that feature for that specific application. A dot skewed towards the left – a negative value –  increases the probability of default. While a dot skewed to the right – a positive value – decreases the probability of default. By calculating and plotting the  Evispot Score for each variable and for each loan it is possible to calculate and visualise the overall model performance. Evispot Traits makes it possible to granularly understand how the AI-model reasons and how it determines the risk of each loan application.

 

Figure 2: Illustrating three variables that the model is using for predicting. Please note that the models that a full overview of the model’s Evispot score is listed at the end of this post.

 

 

By looking closer to the variable Annual income, it can be seen that the AI-model has seen four distinguished intervals. Missing value, below 34800, between 34800 – 218000 and above 218000. By looking at the Evispot score it can be clearly seen that applicants with missing value or less annual income than 34800 are defined as the riskiest group, while above 34800 is seen as a less risky group. By plotting Evispot Score it is possible to visually identify the models characteristic. By using this method it is possible to understand how age, salary or other variables correlate with risk. It helps us to better understand and trust the model performance and robustness.

 

 

Understanding Each Decision

When a new applicant enters, we calculate the Evispot score for this applicant and thereby give you as a creditor the possibility to truly understand the model and the reasoning behind each decision.

Figure 3: Illustrating how the model has reasoned when predicting a new application (red dots), which helps us understand why the model predicted the way it did.

By using the same example as used in figure 2, a comparison between the incoming applicant with the historical applicants can be made. By plotting the new application with red dots it can be seen that this applicant’s Annual income is skewed towards high risk. The applicant’s  Account Balance, as well as Delinquencies, are skewed to low risk. The Evispot score of every variable gives us a complete picture of how and why the model predicts as it does. All in all, the benefits of Evispot Traits allows you as a creditor to gain the benefits of AI in your credit decision scoring while still meeting the demands of transparency and explainability.

 

Interested in learning more about Evispot Traits?
Please contact sebastian.larsson@evispot.ai or call +46(0)70-2332335

 The Model’s Characteristics and Evispot Score for all Variables

read more
Tomas SelldenEvispot Traits – Transparent and Explainable Artificial Intelligence
Presentation-Evispot-1.svg

The Reason Artificial Intelligence is the Future of Lending

 

Lending is built on trust

Lending has always been highly driven by personal relationships and the face-to-face meetings – where trust is key to understand if and how a consumer will pay back his/her debt. The trust is traditionally based on a personal relationship between the applicant and the bank clerk – they have gotten to know each other through several meetings at the local bank office. The digitalisation has truly changed how and when consumers are able to apply for a loan and thereby the requirements on the lenders – the consumers do not have the patience to fill paper forms or to wait days for approval. The consumers want a smooth application process and an answer within a matter of seconds. The lender needs to calculate and understand if the debt will be paid back – and thereby understanding if they can trust the borrower in a matter of a second. This blog post we discuss the transformation that lenders will need to go through in order to keep the pace with the consumers and still being confident that they lend to the right people.

 

Transforming the banking experience

Lenders who haven’t managed to transform into a fully digital and automated application process will have another problem – they will lose customers to the players who have. Based on a survey by Signicat this was obvious – more than one-third of the respondents (consumers) said that they abandoned their loan application as a result of the length and effort required to complete it. Where the requirement of physical papers and waiting-time was the major reason why they abandoned. When consumers come across a frustrating on-boarding will the low-risk and high-risk consumers behaviour differ. Low-risk consumers with good credit who are likely to repay their loans can easily go to another lender. While the high-risk consumers are more willing to jump through challenging on-boarding because they have no other options. Hence, a smooth application process is key for many reasons.

The digitalisation has resulted in several obvious changes when looking to lenders’ way of doing business. During the recent years has the way of marketing changed – going from traditional mails to email campaigns, affiliate networks, agents and social media. Applying for a loan has moved from face-to-face meetings at the bank to become online and mobile. Being able to follow the consumers and keep them satisfied will require the backend of the bank to equally transform. 

Figure 1: The transformation of marketing, application processes and credit underwriting in banking.

The financial companies won’t able to understand all possible customers using the traditional methods. By being able to use more sophisticated credit and risk models applied to more data – the data which you have thought about using but never have – is it possible to find more desirable customers. This can help to reach and understand the ones who were traditionally difficult to score, such as college students and millennials. And since model developed with Artificial Intelligence (AI)  come with automated updates is it possible for lenders to meet changes in the market conditions, new channels (such as mobile) or when the customer populations are reshaping.

 

 

Leveraging all data

Today are we acquiring data at a faster rate than ever. A deeper embrace of the AI and machine learning being used elsewhere at the bank, combining with the rich data being harvested, can meaningfully change a financial firm’s profits for good. 

Yet the widespread use of AI to more sophisticated predict credit risk has eluded the industry until today. The early adopters o AI in credit have measured financial performance benefits – through higher conversion rates or lower credit losses, or a blend of both.

Down below is an illustration of some of the effects going from traditional credit underwriting methods to more sophisticated ones.

Figure 2: The effects and benefits of going from traditional credit models to ones based on machine learning and artificial intelligence. 

 

Fewer resources and more relevance

It can take up to several months to develop, validate, and deploy a new scorecard model. That’s an eternity in the fast-paced digital industry we are living in today. Customer populations, economic conditions, and markets can change dramatically. This could potentially cause introducing errors or human bias into the original model – decreasing its effectiveness. Using AI allows lenders to put powerful models into production quickly and responsibly. Then it is possible to quickly refit new models to adapt to these changing conditions.

 

Full Explainability


AI algorithms are considered a “black box” where models are unable to explain the reasons for its outputs. We at Evispot has recently developed a software which allows us to fully understand the reasoning and decisions of the AI models. This enables lenders to understand the specifics. Explainable AI models help lenders know exactly why applicants are approved or denied, in alignment with their existing business processes and procedures. (A further introduction of explainable AI-models in credits will be featured in an upcoming blog post).

 

We think it is obvious the time for AI in credit decisions is now – do you agree?
Please reach out to sebastian.larsson@evispot.ai to continue the discussion.

read more
Tomas SelldenThe Reason Artificial Intelligence is the Future of Lending

join-our-family-as-a-software-engineer

Join our family as a Software Engineer

As of today, we expect banks to provide all their services digitally, we expect to have the possibility to apply for a loan and know if we are granted a loan or not, within seconds, using our phones whenever it suits us. The traditional credit assessment processes won´t be efficient enough for this transformation, and that is why Evispot exists.

We are a group of technology interested people with competence and experience within credit assessment, finance and deep knowledge within machine learning. Our main goal is to ensure that the credit decisions of tomorrow is given in a transparent, ethical and fair manner without compromising the accuracy.

Work at Evispot

If you choose to get on board with us, you will become part of a growing family, where everyone has big responsibilities and the possibility to to make a big difference. It is our core belief that WE are stronger than I. Therefore, we constantly work with common team-goals, where we together find strategic solutions to build the best possible product to help our customers.

As a software engineer at Evispot, you will be part of building and implementing our AI-platform. Among other things you will be part of implementing solutions for the following challenges:

  • Build and design pipelines and software behind the data flows that exist and soon will come
  • Build data-pipelines which can process thousands of features every second
  • Process and structure unstructured data
  • Build integrations to customer through REST API
  • Work close to our machine learning team and together continuously develop our AI-platform

Who you are?

Most importantly, you are interested in technology and handle large amounts of data. You like to, together with other people, find solutions for new challenges. Furthermore, we appreciate if you previously have been involved in projects where data from different sources have been processed and made interpretable for analysis and model development.

Our primary tool is python, we use it for implementing REST API:s, data processing and model development. We host data at our own servers and use dockers and kubernetes for orchestration. It is great if you previously have been working with these technologies but definitely NOT a requirement.

If you think you would enjoy working at Evispot and have the possibility to work from Gothenburg –  it would be amazing to meet you for a cup of coffee to learn more about what you would like to do in the future.

For more information contact Tomas Sellden at tomas.sellden@evispot.ai with your application and questions.

read more
Tomas Selldenjoin-our-family-as-a-software-engineer

We are hiring – Machine Learning Engineer

WE

Evispot provides the financial industry with the decisions they need in the future – today. We do it by exploring artificial intelligence – because no human, neither configured software can handle the complexity of data that must be evaluated.

 

With 15 years of experience in the credit industry and a strong domain technology interest, we realized the opportunities AI entails in credit decisions. Traditional decisions models are based on a defined amount of parameters. A defined amount of parameters entail limited answers. The possibility to identify patterns, anomalies and multidimensional relationship has not been possible before – until now. The era of traditional credit models is over. New, constantly optimizing, models are the future.

WE + YOU


Evispot is looking for a machine learning engineer to join the team, to take part in the product- and technology development and creating excellent products for the financial industry. You will be a part of disrupting the financial industry and creating next generation credit decisions.

– Take part of Evispot current technology- and product development.
– Develop solutions for real world, large scale problems.
– Hands-on prototyping new algorithms, evaluate with experiments.
– Also, productionize solutions at scale or/and plan for scaling.
– Help drive optimization, testing and tooling to improve data quality.
– Take part of an active start-up who constantly strives to create excellent products.
– Work closely and collaborate with Evispot’s advisors with over 10 years of AI experience.
– Evispot’s office is located in Gothenburg, Sweden.
– Flexible hours and all other startup-related clichés you might desire.

YOU

– Background and experience in machine learning or a related field.
Experience of feature engineering and prototyping machine learning applications, that have been deployed to production
– You are well-versed in programming and scripting (not only R and Matlab)
– Experience implementing machine learning systems in Python and R.
– You care about agile software processes, data-driven development, reliability, and disciplined experimentation
– Grit, and a true problem-solving mindset
– You are excited about joining a startup environment where anything & everything is happening at once.

 

Please contact Tomas Sellden at tomas.sellden@evispot.ai with your application and questions.

read more
Tomas SelldenWe are hiring – Machine Learning Engineer

Machine Learning – an Overview of Algorithms

Machine learning and Artificial intelligence are broad terms, something we often stress when we talk and meet collaborators in the industry. The concept machine learning consists of many subcategories of algorithms, where each category has their own pros and cons making them suitable for different tasks.

To visualize these categories and to exemplify which algorithms are suitable for which task has Evispot created an algorithm fact sheet. The purpose is it to give you an overview of the different subcategories of machine learning.This is a simplified overview where details have been left out and some simplifications have been made. Combinations of algorithms can also be used, this is called ensemble models.

However, the first question to ask yourself before initiating a project is if the amounts of data available. The amount is important to make sure that patterns and insights in data are representative.Machine learning algorithms will always find patterns in the given data (if there is one).  However, the question is if these patterns are representative of the population you want to analyze.

Evispot most commonly used machine learning algorithms.

Evispot’s algorithm sheet.

This is merely an overview of the most common algorithms used by Evispot, and our view of them.

 

Anomaly Detection Algorithms

Anomaly detection algorithms are used for identification of items, events, and observations which don’t follow expected patterns in the given dataset. As shown in the fact sheet above, one category is rare if the data. The algorithms are suitable for tasks where the majority of the observations are similar and a few differences from the masses. The typical anomaly algorithms are used to identify issues or problems as bank frauds, errors in a text or medical problems. Anomalies are also referred to as outliers, novelties, noise, deviations, and exceptions. Example of anomaly detection algorithms is Support Vector Machines (SVM), PCA-based anomaly and Cluster analysis-based outlier detection.

However, the interesting objects are often not rare objects, but unexpected bursts in activity. This pattern does not adhere to the common statistical definition of an outlier as a rare object, and many outlier detection methods (in particular unsupervised methods) will fail on such data unless it has been aggregated appropriately. Instead, an anomaly detection algorithm may be able to detect the microclusters formed by these patterns.

Example of usage: Identifying bank frauds, errors in text and medical problems.


Classification Algorithms

Classifications algorithms are used to identify where, to a set of categories, a new observation belongs to. This is done on the basis of the training set of data containing observations whose category is known. Classification algorithms are used for patterns recognition. As can be seen in the figure are neural networks and random forest algorithms both included in the category of classification. As the figure shows, one difference between these algorithms is the transparency of them. Where neural network, most often, retain higher accuracy than random forest but lack in terms of transparency, making them suitable for different kinds of tasks.

Example of usage: Categorising loan applications to a given number of risk categories and classifying images based on whats on them.

 

Clustering Algorithms

Clustering algorithms are used to group a set of objects in such way that the objects in the same group (defined as clusters) are more similar to each other than to those in the other groups. Simply put – grouping many objects to clusters containing similar objects. The notion of a “cluster” cannot be precisely defined which is one of the reasons why there are so many clustering algorithms. In difference to classifications, algorithms aren’t clustering algorithms suitable for tasks with labels data and learning by examples. Clustering is instead used to find the connections in the dataset.

Cluster analysis as such is not an automatic task, but an iterative process of knowledge discovery or interactive multi-objective optimization that involves trial and failure. It is often necessary to modify data preprocessing and model parameters until the result achieves the desired properties.Example of suitable clustering algorithms includes KMeans, Spectral Clustering, and Agglomerative Clustering.

Example of usage: Identifying payment archetypes based payment behaviours, identifying groups of genrés in large groups of articles. 

 

Regression Algorithms

When it comes to credit decisions regression algorithms are commonly known, since many of the existing credit scoring models have this is basis. Regression analysis is widely used for prediction and forecasting, where its use has substantial overlap with the field of machine learning. Regression algorithms are used to estimate real values based on a continuous variable(s). The models establish a relationship between independent and dependent variables by fitting the best line. This best fit line is known as regression line and represented by a linear equation Y= a *X + b.

The best way to understand linear regression is to relive this experience of childhood. Let us say, you ask a child in fifth grade to arrange people in his class by increasing order of weight, without asking them their weights! What do you think the child will do? He/she would likely look (visually analyze) at the height and build of people and arrange them using a combination of these visible parameters. This is a linear regression in real life. The child has actually figured out that height and build would be correlated to the weight by a relationship, which looks like the equation above

Examples of usage: Forecasting value of a house

 

Time Series Algorithms

Time Series algorithms uses a set of powerful statistical and machine learning tools for predicting future events based on past data. Most commonly, a time series is a sequence taken at successive equally spaced points in time. By indexing events on a timeline is it possible to forecast new events. This can be used to forecast when an invoice will be paid or forecasting a cash flow based on accounting data. Time series algorithms include HMM discrete and AR continuous.

Example of usage: Forecasting cash flow on historical accounting data.

 

Endnote

As earlier mentioned, this algorithm sheet is used to give you a holistic overview of which algorithms we most commonly use and their characteristics.

These can be combined and each category contains several other algorithms. If you have any questions regarding algorithms or machine learning in general, don’t hesitate to contact us on hello@evispot.ai

For more information on machine learning, please visit:
https://en.wikipedia.org/wiki/Outline_of_machine_learning
https://machinelearningmastery.com/a-tour-of-machine-learning-algorithms/
https://www.kdnuggets.com/2016/08/10-algorithms-machine-learning-engineers.html

 

read more
Tomas SelldenMachine Learning – an Overview of Algorithms

Evispot Articles: Case #1 – Detecting ​​Default ​​Payments ​​of ​​Credit ​​Card ​​Clients using Machine Learning

Hope you have had a great summer! While many have been lying on the beach and enjoyed vacation, Evispot has worked on interesting machine learning cases. 

During the upcoming weeks we will share our experiences in a serie of articles. The idea is to give you insights and a deeper understanding of machine learning in credit scoring. The articles follow Evispot´s development process going from raw data to actionable insights – and our hope is that you will share our view: 

Machine learning is the future in credit scoring

Case#1 –  Detecting ​​Default ​​Payments ​​of ​​Credit ​​Card ​​Clients using Machine Learning

The first article is about predicting default payments of credit card clients, using demographic data, credit data and history of payment bill statements of credit card clients. In this example, we use logistic regression to briefly demonstrate a classification problem, and how classification can be more useful than just predicting the outcome variable.

Follow the link to get to the article:
Case #1 – Detecting ​​Default ​​Payments ​​of ​​Credit ​​Card ​​Clients using Machine Learning 

If you have any questions or data you want to explore feel free to contact:

Sebastian.Larsson@evispot.ai, CEO & Co-founder Evispot

 

read more
Tomas SelldenEvispot Articles: Case #1 – Detecting ​​Default ​​Payments ​​of ​​Credit ​​Card ​​Clients using Machine Learning