ANALYTICS, MACHINE LEARNING AND AI
Success of any derived business insight is critical to the quality of data, sample and size of data, understanding of the business domain, applied logic, applied statistical or mathematical analysis for derived outcomes, modeling, training, operationalization, correction and re-training the model, again operationalizing and deriving results. As simple as it may sound, years of effort, have today given rise to manageable processes, which is today classified as part of Enterprise AI. Redsilver Technologies with a strong focus and belief in offering its customers true value in terms of, better business insights and improved business process works with some of the leaders in the field of Enterprise AI, offering its customers tested and world recognized solutions.
We offer solutions that cover a wide array of applications and solutions combining to help drive business growth, competitive edge and business intelligence while improving operational efficiency and predicting business outcomes. We help and address the needs of team members like business analysts, data architects, data engineers, data scientists, management and stake holders.
Our solutions span the following broad Technology and Solutions domains :
DATA ANALYSIS
The core and crux of all data analysis be it using Statistical data analysis techniques, Machine Learning or Deep Learning/Artificial Intelligence methods are as successful as per the quality of data being analyzed. Data sources across an organization are plenty from Big Data stores to the last mile Excel and CSV file formats.
Being able to analyze the data in real time and using the right data cleaning techniques to cleanse missing data, null records, inconsistencies in the data in terms of date formats, missing numericals/numbers, data duplication and deduping the data, data entry errors, data noise or harmonics in signal data analysis etc. are the many kinds of data errors that occur every single time that we pool data from various sources.
The need of the hour is a comprehensive data cleansing solution which combines seamless aggregation of data from multiple data sources, detecting.
-
Connect to any Data Source - SQL Databases, NoSQL Databases, Hadoop and Spark supported distributions, Hadoop File formats, Remote data sources, Cloud Object Storage, Custom Data Sources (through Rest API)
-
Automatically detect dataset schema and data types from all your existing connections
-
Scale your transformations by running them directly in distributed computations systems (SQL, Hive, Spark, Impala)
-
Assign semantic meanings to your datasets columns
-
Build univariate statistics automatically & derive data quality checks
-
Dataset audit and Automatically produce data quality and statistical analysis of entire Dataiku datasets
-
Data Transformation - Design your data transformation jobs using a point-and-click interface, select from Group, Filter, Sort, Stack, Join, Window, Sync, Distinct, Top-N, Pivot, Split etc.
MACHINE LEARNING
Dataiku offers the latest machine learning technologies all in one place so that data scientists can focus on what they do best: building and optimizing the right model for the use case at hand.
Automated ML strategies:
-
Quick prototypes, Interpretable models, High performance features handling for Machine Learning.
-
Support for numerical, categorical, text and vector features.
-
Automatic preprocessing of categorical features (Dummy encoding, impact coding, hashing, custom preprocessing, etc.).
-
Automatic preprocessing of numerical features (Standard scaling, quantile-based binning, custom preprocessing, etc.).
-
Automatic preprocessing of text features (TF/IDF, Hashing trick, Truncated SVD, Custom preprocessing)
-
Various missing values imputation strategies
-
Choose between several ML backends to train your models - TensorFlow, Keras, Scikit-learn, XGBoost, MLLib (Logistic Regression, Linear Regression, Decision Trees, Random Forest, Gradient Boosted Trees, Naive Bayes, Custom models), H20 based (Deep Learning, GBM, GLM, Random Forest, Native Bayes etc.).
-
Model Deployment - Model versioning, Batch scoring, Real-time scoring.
-
Expose your models through REST API’s for realtime scoring by other applications
ARTIFICIAL INTELLIGENCE
Algorithms make a lot of decisions that impact society without us knowing. Algorithms decide which prisoners are most likely to commit crimes once they are released into society, or who is more likely to commit crime or who are the rioters at a scene using facial mapping or who is happy from facial contours etc. But, transparency to the process of decision making or algorithms is known to few, can be challenged or can change over a period. What these algorithms are doing is also potentially taking away part of what makes us human. Our right to make decisions is one of those things. Dataiku eases most of this decision making, traceability, governance, model training, deployment etc.
Deep Learning
-
Support for Keras with TensorFlow backend
-
User-defined model architecture
-
Personalize training settings
-
Support for multiple inputs for your models
-
Support for CPU and GPU
-
Support pre-trained models
-
Extract features from images
-
Tensor board integration
Unsupervised Learning
-
Automated features engineering (similar to Supervised learning)
-
Optional dimensionality reduction
-
Outliers detection
-
Algorithms - K-means, Gaussian Mixture, Agglomerative Clustering, Spectral Clustering, DBSCAN, Interactive Clustering (Two-step clustering), Isolation Forest (Anomaly Detection), Custom Models
-
Train models over Kubernetes
-
Automation Workflows
-
Partitioning - Leverage HDFS or SQL partitioning mechanisms to optimize computation time
-
Collaboration, Coding,
-
Governance and Security, Model Bias etc. and many more such features and functionality