Supervised Techniques are very simple to deal with. Hence unsupervised technique is the method when we do not know anything about the dataset. Clustering tends to patterns and sometimes to deal with outliers. Data mining is the process of looking at the patterns in a dataset. Clustering plays its role to see the natural behaviors or observations.
New Technology like deep learning (Neural Networks) that acts like a human brain also introduced classification models. Convolutional Neural Network (CNN) model classifies the images, does clustering can do so? Clustering is a machine learning model that cannot compete with deep learning. But deep learning can work far better other than ML model image datasets.
ML classification like clustering is good, it depends on what you want or what are you doing with the dataset.
Disadvantages of Clustering in Machine Learning
Large Dataset
Clustering cannot work with large datasets, grouping the data with similar properties does not give the same results every time.
Cannot Work With Image Dataser Properly
Images have pixels, that combine in rows and columns to make a pattern. Edges of the image are those points where clustering cannot find out a similar group of the dataset.
Unable To Work With Outlier
K-mean is a type of clustering that fails to look into the outliers of a dataset. For example, 45, 99, 41, 44, 49, 47, 3, 43, 42. In this series of data, 3 and 99 are outliers that are unable to reach out. K-mean calculates the average of near values while outliers come into it and fail.
Deal With Imbalanced Data
Preprocessing is a data mining technique that helps in training the model. The dataset is for training and testing. It is very important to understand that clustring is unable to resolve the issues of dataset.