Special Offer - Enroll Now and Get 2 Course at ₹25000/- Only Explore Now!

All Courses
Machine Learning with Python Interview Questions and Answers

Machine Learning with Python Interview Questions and Answers

April 13th, 2019

In case you’re searching for Machine Learning with Python Interview Questions and answers for Experienced or Freshers, you are at the correct place. There is parcel of chances from many presumed organizations on the planet. The Machine Learning with Python advertise is relied upon to develop to more than $5 billion by 2020, from just $180 million, as per Machine Learning with Python industry gauges. In this way, despite everything you have the chance to push forward in your vocation in Machine Learning with Python Development. GangBoard offers Advanced Machine Learning with Python Interview Questions and answers that assist you in splitting your Machine Learning with Python interview and procure dream vocation as Machine Learning with Python Developer.

Best Machine Learning with Python Interview Questions and Answers

Do you believe that you have the right stuff to be a section in the advancement of future Machine Learning with Python , the GangBoard is here to control you to sustain your vocation. Various fortune 1000 organizations around the world are utilizing the innovation of Machine Learning with Python to meet the necessities of their customers. Machine Learning with Python is being utilized as a part of numerous businesses. To have a great development in Machine Learning with Python work, our page furnishes you with nitty-gritty data as Machine Learning with Python prospective employee meeting questions and answers. Machine Learning with Python Interview Questions and answers are prepared by 10+ years experienced industry experts. Machine Learning with Python Interview Questions and answers are very useful to the Fresher or Experienced person who is looking for the new challenging job from the reputed company. Our Machine Learning with Python Questions and answers are very simple and have more examples for your better understanding.

By this Machine Learning with Python Interview Questions and answers, many students are got placed in many reputed companies with high package salary. So utilize our Machine Learning with Python Interview Questions and answers to grow in your career.

Q1) How to Assign Code to the List?

Answer: Using this syntax continuation, we can assign symbolic value to any list.
Mylist = [None] * 10 (none of the 10’s list)

Q2) Give me two important tasks in the pants?


  • Series
  • Data frame

Q3) What is the difference between iloc and loc activity?


  • Take the pieces based on the lock labels (features).
  • It uses the Index based position.

Q4) What package is used to import data from the Oracle server?

Answer: We use CX_Oracle modules to link Python with Oracle server.

Q5) Import of Flat File / CSV in Baidan


  • Read_csv
  • Generatorrext

Q6) How to read an Excel file without a file file in the Byndah?

Answer: Read the Excel file using the Xlsreader module and manipulate it.

Q7) What do the review process do?

Answer: You can change the data without changing the data.

Q8) What do Dummies do?

Answer: It can alter the duplicate / cursor variables alternately.

Q9) What are two types of polymorphism?


  • Time polymorphism / method overloading compilation
  • Run time, Polymorphism / Mode

Q10) What are Lambda Functions in Python?

Answer: Python, anonymous function is a function defined without a name. When normal functions are defined using a defined keyword, Python is defined as anonymous functions using the Lambda word. Therefore, anonymous functions are called Lambda functions.

Q11) When to use the yield instead of returning to the crazy?

Answer: Performance Reports pauses the functionality of the activity, returns the caller back to the caller, but retains enough condition to activate the function and resumes where it is left. Once the restart is done, the yield starts functioning, and when the yield starts running. It allows its code to produce continuous values over time, but they simultaneously calculate them and send them a list.

Q12) What are Generators in Python?

Answer: Byrne Generators This is a simple way of creating platforms. When simply speaking, a generator is a substance that represents an object (reboot), and we can re-run it (at a time).
Data pre-processing interview questions

Q13) How can I retrieve an important part of data collection?

Answer: The distance from the remaining studies is limited to the limited violations. As a result, they can be flexible or disagreeable for any analysis in any analysis in the database. It is therefore important to detect and be harmful enough.
When a 100% reassurance is due to a test / transcription / etc error, they should only be rejected if they are exited. Otherwise, the removal of the outlines would have been underestimated.

Q14) Difference between distinct, bivariate and multivariate analysis?

Answer: Analysis Data Simple analytics analysis of data analysis that contains only one variable. This is a variable because it does not cope with the causes or relationships. The main purpose of the unique analysis is to describe data and discover the forms inside it. Ex. Average, method, intermediate, range, variance, max, at least, quartz and standard deviation
Bivariate Analysis is used to find out if there is a relationship between two different variables.
Ex. Skater Plate, Co
Multidimensional analysis is analysis of three or more variables. There are a number of ways to analyze diversity according to your goals.
Ex. Cluster Analysis, Multiple Recreation Analysis

Q15) What is the curtain?

Answer: In many setback analysis, one of the forecasts is in contrast to the other predictor / dependent, then this problem is known as collinearity.

Q16) Why is a useful metric?

Answer: Can you measure a size from one person? Relationship (but some of the below examples can not be as often as we can see) can refer to the presence of a causal relationship. Many modeling techniques are used as a base size and base combination

Q17) What is the forward selection of data pre-processing?

Answer: The perspective option is a rectification method that begins without any aspect of the model. In each iteration, we will add a better way to improve our model until we add a new variable to improve the performance of the model.

Q18) What is the removal of data backwards in advance?

Answer: Disadvantage eliminates at least every significant aspect of each reaction that starts with all the features and improves the performance of the model. We do this again until there is no improvement in removing features.

Q19) What is the removal of the recursive feature in data pre-processing?

Answer: This is a greedy optimization algorithm that finds a good style feature subset. It creates repetitive models and each reboot keeps aside the best or worse performance feature. This creates the next model with the left features until all features are exhausted. Then there will be elements based on the order to remove them.

Q20) What is Unequal Data?

Answer: Balanced data sets for classification issues are special classes, and class distribution between classes is not uniform. In general, they are created by two types: the majority (negative) class and minority (positive) class.
Because this type of set data mining is a new challenging problem, because standard classification protocols typically considers a consistent training package and most of it is thinking of a pro in class.

Q21) What is the Cross Verification?

Answer: Cross Verification is a recursive procedure used to evaluate machine learning models in a particular data model.

Q22) The performance of the K modular system?

Answer: The general procedure is as follows:

  • Rotate the database randomly.
  • Divide databases into k groups
  • For each individual group:
  • Take out as a group or take the test data package
  • Take the remaining groups as training data set
  • Apply a sample on training and evaluate the test package
  • Retain valuation value and reject the model

Shorten the model’s ability to model the model’s rating scores

Q23) What is the nature of the data signal?

Answer: Data normalization is the process for restructuring one or more attributes between 0 to 1. This means that the largest value for each attribute is 1 and smaller. You know that distribution or distribution of your data is not a cache (bell curve).

Q24) What is standardization?

Answer: Data rate evaluation is the process of restructuring one or more attributes, whereby they are the average value of the 0 and the standard nomination 1. Standarty considers your data to be a gauge (bell curve) distribution. This will not be true, but your attribute distribution is a very effective technique
Machine Learning Interview Questions

Q25) What are the three stages for creating a model in machine learning?


  • Model building
  • Model test
  • Applying the model

Q26) Keep in mind that you are working in a data system, and explain whether you choose key variables.

Answer: Some methods are used to select the following critical variables:

  • Using the loso regression system.
  • Using the Random Forest, the plot variable imprtance chart.
  • Using linear lag.

Q27) Why are the innocent demons ‘innocent’?

Answer: Since innocent ghosts are very ‘naïve’, all aspects of the data set are equally important and independent. As we know, this assumption is rare in the real world situation.

Q28) How is KI different?

Answer: K-Recent neighboring countries have a classification algorithm, while k-object is an uncontrolled clustering algorithm. Although the mechanisms seem to look the same, you need data that you need to classify an unnamed point (neighboring area) to work with neighboring neighboring countries. K-material clustering requires only a single point of reference and a starting point: Algorithm can learn how to group the group into groups by taking unstoppable points and calculating the gap between different points.
The significant difference here is that the KNN has to be named for points, which require supervised learning, while the k-object does not – there is no supervision.

Q29) Is It the Most Important For You Model Model Accuracy or Model Performance?

Answer: This question tests your grip on the machine learning model performance nuances! Machine Learning Interview Questions are often headed towards the details. There are models with greater accuracy, which advance the power of the advance – how is it realized?
Well, model accuracy model performance is only a subset of how to do it, sometimes it’s a misguided guide. For example, if you find millions of models in a large database, if only a very small number of fraud cases, the most accurate model does not contradict any fraud. However, it will be ineffective in advance – insisting that there is no fraud on a model designed to detect fraud! Questions like these help you to demonstrate that you need to understand the model’s accuracy.

Q30) When Should You Use Taxonomy on Retreat?

Answer: Sorting creates a database for distinct values and strict categories, while you record the conclusions that allow you to distinguish the difference between individual points. You can categorize the consequences if you want to reflect the combination of data points in your database for certain specific sections. (For example, female names, when compared to male, female, male and female).

Q31) What is upwards?

Answer: Overfitting occurs when a statistical model or machine learning algorithm captures data noise. Intuitively, overfitting occurs when the model or algorithm data fits very well. In particular, if a sample or algorithm is showing low mumps, there is a high variation. Floating is often a result of a more complex model, and it is compatible with many sample samples and test data to compare their predictive accuracy using a validation or cross-estimate.

Q32) What is downwards?

Answer: Underfitting occurs when a statistical model or machine learning algorithm does not catch the basic trend of data. Instinctively, if the sample or algorithm does not match the data correctly, it shows the high independence, especially if it has shown a sample or algorithmic variance. The foundation is often a very simple model result.

Q33) How do you make sure that you do not block a model?

Answer: It is a simple problem with a basic problem with machine learning: training data is likely to carry that data noise through overtitting and testing package, thus providing inaccurate generalization.

Q34) What are the main guidelines to avoid excesses?


  • Simplify the sample: You can reduce the transition by lower variables and parameters, thus eliminating some of the noise in training data.
  • Use k-folds cross-validation for cross-checking techniques.
  • regulatory techniques such as LASOO, which are some sample parameters to be punished if they make the tablet.

Q35) How to handle unbalanced databases?

Answer: When you have an unbalanced database, for example, a classification test and 90% of data is in a class. This leads to problems: if there is no computing power in the other section of data data, 90%.

Q36) What is Learning Strength?

Answer: Reinforcement learning is a type of machine learning, and thus a branch of artificial intelligence. In order to increase its performance, it allows machines and software agents to automatically determine the best possible performance in a given environment. The simple reward idea for the agent to learn its behavior is essential; This is known as the Reinforcement Signal.
One fact is, the reinforcement learning is defined by a particular type of problem, and all its solutions are classified as reinforcement learning algorithms. The problem is, an agent must decide on the basis of his current state and decide the best action. When this step is repeated, the problem is called Marcov Decision Making.

Q37) What is the result tree?

Answer: A conclusion is a concrete representation for all solutions that are based on specific conditions. It starts with a single box (or root), just like a tree, because it gives a solution like a tree.

Q38) What is a random forest?

Answer: Random forest produces many end-results trees and merges them to get more accurate and consistent predictions.

Q39) What is the central trend?

Answer: The central trend is a value that attempts to describe the data set by identifying the position of the central within a set of measurement data. Therefore, the activities of the central tendencies are sometimes called central location operations. They are categorized as abstract statistics.
Example: average, average, pattern

Q40) When we use Pearson’s relationship co-efficient method?

Answer: Pearson communicates the linear relationship between two consecutive variables involved. Relationship linear is when the change in a variable is related to a proportional change in the other variable.
For example, a Pearson contact can be used to assess whether the increase in the temperature of your production facilities is associated with lower thickness of your chocolate coatings.

Q41) What is the standard deviation, how is it calculated?

Answer: Standard Disadvantage (SD) is a statistical measure, which captures the meanings of the meanings and rankings.
Step 1: Find the average.
Step 2: Find the average square of its distance for each data point.
Step 3: A total of values from step 2.
Step 4: Separate the number of data points.
Step 5: Take a square hunt.

Q42) What is Z Score?

Answer: The z-score is the standard distortion count from a data point on average. But technically this is a source of how many constant changes are above or above the population. A z-score is known as a fixed value and can be placed in a normal distribution ramp. It eliminates values from the database that are lower than Z times 3 times.

Q43) What is Type I and Type II Error?

Type I Error: A Type I error occurs when a null hypothesis rejects the researcher. The probability of performing a type I error is called a significance, and is often denoted by α.
Type II Error: When a researcher accepts a null hypothesis wrong, Type II error occurs. The probability that a type II error occurs is called beta, and is often denoted by β. The probability of a Type II error is called Power Test.

Q44) What is the remainder?

Answer: In the review analysis, the difference between the estimated value of the dependent variable (y) and the calculated value (y) is called the remainder (d). Every data point is a remainder.
Remaining = Value Value – Estimated value e = y – y
The total and the remaining remaining are equal to zero. Σ e = 0 and e = 0.

Q45) What is a Sample Model Test?

Answer: A sample T-test is used to check whether the population mean is significantly different from the value of some hypotheses.

Q46) What is F Statistics?

Answer: If you have a significant difference in the way between the two people you will find an FO point of value when you are running an ANOVA test or a regression analysis. This is just like a T-test a D statistic; If the A-T test is a variable statistically significant and will tell you if a F test variable is of significant significance.

Q47) What is ANOVA?

Answer: ANOVA is used for comparison with three or more models.

  • One way is ANOVA (which is an independent variable).
  • Two way ANOVA (there are two distinct variables)

Q48) What is data preprocessing in machine learning in python?


  • Pre- management is mention as the changes are activated to the facts before giving it to an algorithm.
  • Data preprocessing is a method used to change the raw data in a clean data group. Data is collected from various origins and gather in basic format is not practical for examining.
  • To get the best outcomes from the registered model in the projects of machine learning and the pattern of the facts should be well arranged.

Q49) What is a statically method? Does it use?

Answer: It is known as Hypothesis testing that is used to take for the agreement of experimental facts for true for the whole population or not. It is an acceptance of the parameters of the population. It is a trial to control the connection between two data groups.
It is very important for a method in statics. It is for assessing two physical full declarations of the population to examine and tell the supreme help of the sample data. Searching a statically remark is a hypothesis test. There are two terms normalization and standard normalization.

Q50) What the parameters of hypothesis testing?

Answer: Null hypotheses – It is known as a general statement or default position of no connection between two regular events in presumed statics. It is also a primary acceptance.
The alternative hypothesis – It is used for analyzing the outcome of a real effect. It is used for testing hypotheses opposite to Null hypothesis. It shapes the community is a small, great or differs from the principle of hypothesis in the null hypothesis.

Q51) What is a business dataset?

Answer: It is known as categorical data used in machine learning with python.
For example – Customers are commonly defined from the country, gender, age, name, etc. and the commodity is also defined by the type of product, producer, vendor, etc. It is very easy for people and difficult for the algorithm of machine learning because of various cause

  • Mostly machine models are in algebraic
  • ML packages convert class facts into numerical mechanical.
  • Unqualified variables contain a large number of levels to appears as a small number of examples.

Q52) Name the categories of Machine Learning Algorithm with Python?


  • Supervised – In it the feedback is contained to the computer to provide for the trial data for learning. The system manages the sample inputs and needed the output to learn a common rule to measure inputs to outputs.
  • Unsupervised – No tag is obtained by the python machine learning algorithm. Only a group of inputs is provided. It depends on itself to search the construction in the input. This is considered as the achievement for future learning and can analyze unsupervised learning as clustering, association.

Q53) Explain Anova?

Answer: It is known as statistical hypothesis testing for examining exploratory facts, the outcome of the experiment is known as statically significant when it occurs by chance and presuming the fact of a null hypothesis. If the p-value is less than the approach level then it examines the refusal of the null hypothesis but it occurs only if the prior value of the null hypothesis is less. All the set of the null hypothesis is instance examples of a similar population.

Q54) Mention the reason for python is the best for machine learning?

Answer: Python is a very convenient programming language for research and development in the field of machine learning. It has ML languages such as R, JAVA, Scala, Julia, etc. straggle behind.

  • It is very simple and readable for both developers and exploratory students. It permits us to finish the project without using more codes.
  • Python contains various and numerous libraries and frameworks so that we can save our time. Libraries such as Keras, TensorFlow, Scikit-learn.
  • It is portable, extensible and to support community and corporate.

Q55) What is scikit learning?

Answer: It is considering as an open-source of python library to apply a wide variety of machine learning, crop-validation, visualization algorithm, pre-processing by the use of the combination.

  • An effective and simple implementation for data mining and to examine data.
  • Available and renewable for everyone for different contexts.
  • It is constructed don the top of Numpy, SciPy, and used commercially.

Q56) Define the uses of PCA?


  • It is for searching inner connectivity in the middle of variables of data It is used for explaining and envision data
  • Analyzing become easy and simple when the counting of variable drops
  • It is usually envisioned hereditary distance and applicability between the community.
  • It acts on the square a balanced cast and a natural sum of squares and cross product cast.

Q57) How to compute the dot product two vectors xx and yy?

Answer: With the help of the Kernel trick is also known as the generalized dot product. It engages functions to authorize in a very high proportion space without clarifying the computing co-relation of points in the proposition. It calculates the inside products between the images of the pairs of data. Mostly algorithm is shown in the expression of the inside of the product.

Q58) What does K- means?

Answer: It means gathering an unsupervised machine learning algorithm. To analyze the data without experience in tagging the data. After running the algorithm, the list is described and easy to allot the most suitable list. It contains such as user profiling, segmentation of the market, computer vision, astronomy and search engines.

Q59) What is a type of supervised machine learning algorithm?

Answer: It is effortless to apply in the primary format but it acts difficult for grading the projects. Al the facts are used for at the time of grading a new example. It is known as a non-parametric algorithm that accepts nothing of unlined data. It is considering a lazy learning algorithm as it does not contain any particular training.

Q60) Explain a decision tree pruned?

Answer: It is known as a supervised learning algorithm that is used in grading and backsliding projects. The trees are allotted to the details built on the learning algorithm to use on various estimates of the details achieved from learning. It is used for solving the problem when having not only extent but also an unqualified feature of input and target.

Q61) How to detect fraud in datasets?

Answer: With the help of Model accuracy is a substitute for model performance.
For example – If anyone wants to search fraud from the details which are huge with an example of millions. When the huge opposition of cases is fraud then the high accuracy model will forecast no fraud.

Q62) Name the extension build on formalized linear regression?

Answer: Lasso is an extension having a small twist It is the method to defeat the disadvantages of Ridge regression by penalizing the high principles of the collaborating B, but the fixing to zero if they not suitable. You can finish with few features contains in the model.

Q63) Define Ridge regression?

Answer: It is known as the extension of linear regression to formalized with a linear regression model. By the of the method named cross-validation, the parameter of the scalar can be learned. To apply for collaborating Bas low but it not applied as Zero. It adds a squared magnitude of collaborating as a discipline for losing the function.

Q64)  What are the technique to manage an Imbalanced Dataset?


  • By using the accurate judgment of grade for the models. Choose the example which is suitable.
  • Resample your unbalanced data set by the help of two methods known as under-sampling and over-sampling.
  • To reduce the issue of imbalance data the use K-fold cross-validation perfectly.
  • Keep together various to sample again the datasets.
  • Sample again with the various ratios between the rare and abundant class.
  • Gather all the plentiful class
  • Models to be designed

Q65) How to change consumer evolution?

Answer: It is a decomposition way of using an emotional chart to show the proportion. MDS is used to change consumer evolution into distances parents in the multi-dimensional scaling. It is considered as the experimental technique for evaluating the unknown proportion for the products. To disclose the relative judgment of the products when the fundamentals are connection unknown.

Q66) How to notice heteroscedasticity in a simple regression model?

Answer: There is no heteroscedasticity in the linear regression. The difference in remaining will not increase with suitable principles of the response variable. When the model is constructed is difficult to define some sample in the response variable is viewed in the remaining. As an ineffective and changeable regression model is to surrender the strange forecast.

Q67) Mention the NumPy and Scipy?


  • NumPy – It is used for the primary operation like classifying, listing, fundamental function on the arrangement of a data type. It includes all numeric python and a multi-dimensional array of the item. It is written in C are used in different operation of the facts
  • Scipy – It is known as a Scientific python that includes all the algebraic functions. It helps operations such as integration, differences, grading optimization. It is popular because of its speed. It does not have a group of idea like it is more functional.

Q68) What is T in ML?

Answer: It describes the real-world issue to resolve. The issue is to search the suitable house price in a particular place or to search the best marketing master plan etc.It is hard to resolve ML build tasks at the time it is built on the procedure and system for operating on the facts points.

Q69) What is a quantitive metric?

Answer: Performance(p) informs us about how the model is executing the task with the help of experience(E). There are numerous metrics to aid in explaining the performance of ML such as its accurate score, F1 score, confusion matrix, precision, recall, etc. The calculation informs about the ML algorithm is conducting as a personal requirement.

Q70) What is an experience?

Answer: It is knowledge obtained from the data points given by the model. After getting the details then the model runs and grasps basic design. While constructing comparison with the human being, as a person is obtaining experience from the different a loot such as situation and relationship.

Q71) Define the libraries for machine learning?

Answer: It is a way of programming a computer to grasp the various type of facts. Machine learning is the area of education for provides the system the capacity to grasp without clarity of the program. It is very difficult to resolve different types of issues. The machine learning project is used by physically coding for every algorithm and mathematical and statically equation. Python libraries are Numpy, Theano, Scipy, Scikit-learn, etc.

Q72) What are the filter methods?

Answer: It depends upon the ordinary facility of the data for analyzing and to catch the feature subnet. It does not include any mining algorithm. Filter methods used an accurate evaluation basis that contains distance, details, province, and thickness. It uses the basis of the principle of grading strategy and grading order method for different choices.

Q73) What is used for a greedy search to find a suitable feature subnet?

Answer: With the help of Recursive feature elimination. And generates models and examine the super or bad performing feature to every repetition. It builds the consecutive models with the remaining feature unless every feature is analyzed. After it grades the feature placed on the order of the rejection.

Q74) Define an evolutionary algorithm for feature selection?

Answer: The chromosomes of the creature, influence for getting over the succeeding origination for the best accommodation. The Genetic Algorithm is a heuristic development method attracted by the process of natural expansion. This function is for developing the conclusion action for an auguring model. To decrease the issue of the model on a separate data group does nit generate the model.

Q75) Name the challenges and application of Machine learning


  • Provide low-quality data to generate the issue connected with data processing.
  • It is a very time-consuming task for data acquisition, feature extraction, and retrieval.
  • Absence of expert resources
  • Error of overfitting and underfitting
  • Profanity of dimensionality
  • Problematic in the deployment


  • Analyzing emotions
  • Analyzing sentiments
  • Error Disclosure and avoidance
  • Whether calculating and indicating
  • Fraud analyzing and avoidance

Q76) Define KNN?

Answer: It is very convenient, easy to understand, adaptable is one of the best machine learning algorithms. It is used for different applications like finance, healthcare, political, science, handwriting judgment, image analyzing, video analyzing. It analysis if the loan is safe or risky. It is used for arrangement and backsliding issues.

Q77) How a KNN algorithm performs?

Answer: It is the number of the nearest neighbors and the basis of the finalizing factor. It is an odd number. For example, if the classes are 2 then K=1 the algorithm is the nearest neighbor number or P1 is the point required to conclude. Now search the one nearest point P1 and then the tag of the nearest point allows P1

Q78) What are the bedamn of capacity?

Answer: KNN is good for the low number of features. At the time when the counting of features gets higher than it needs more data. Enhancing the capacity to conduct the issue of overfitting. To required facts is to increase as expanding the number of capacities. We require to act main element detection before covering any machine learning algorithm.

Q79) determine the numbers of neighbors in KNN?

Answer: No excellent number of neighbors is fit for every kind of data group. Every data group has its own needs. A small counting of neighbors then the sound may have a great effect on the outcomes. And on the large counting of neighbors are more easy to adjust with the small preference but big example. And a big counting of neighbors will contain effortless agreement.

Q80) How to enhance KNN?

Answer: To control the data on a similar proportion is high in demand. The controlling the area is known as in the middle of 0 and 1. KNN is perfect for huge proportions of data. In most of the cases, the proportion requires to enhance the activities. It also manages the missing value to assist in the improvement of outcomes.

Q81) How to analyze quantitive data?

Answer: With the help of Univariate analysis, contain one reliable variable and used for experimenting with the hypothesis and to make obstruction. The purpose is to run data, defines and compile the pattern in it. In the group of data, it examines every example independently. It detects the area and the basic movement of the values. Data is dropped under a variable. It is used to detect the facts of cases contains a single variable for a single component in the data pattern.

Q82) Name the data which commits three or more variables?

Answer: It is Multivariate data. When a web developer wishes to analyze the click and alternation grading of the four various pages of web between men and women. Then the connectivity between the variable is mapped by the multivariate variables.
For example- In a medical experiment on the drugs for a=detecting the numerous replies of a patient on a drug.