Determining a health of a network device during onboarding

A device, comprising:

• one or more memories;
• and
• one or more processors to:
• receive key performance indicators (KPIs) corresponding to a plurality of network devices of a network;
• generate features based on the KPIs by generating the features corresponding to the KPIs related to components of the plurality of network devices, wherein the plurality of network devices include different types of network devices;
• select boundaries for a machine learning model based on the KPIs and the features;
• process the KPIs and the features, with the machine learning model and based on the boundaries, to generate steady state data corresponding to the plurality of network devices,
• wherein the machine learning model includes one of a multilayer perceptron model, a long short-term memory (LSTM) model, a bidirectional LSTM model, a Prophet model, or a clustering model, and
• wherein the steady state data is associated with a forecasted range for multiple percentiles in a future time related to the KPIs, that is determined based on using the machine learning model;
• receive new KPIs corresponding to a network device to be added to the network;
• generate new features based on the new KPIs;
• compare, during onboarding of the network device, the steady state data with the new KPIs and the new features to identify one or more anomalies corresponding to the network device being onboarded to be added to the network, wherein comparing the steady state data comprises:
• utilizing for the comparison, a subset of the steady state data corresponding to a subset of particular network devices of a same device type the plurality of network devices as a type of device of the network device,
• wherein the subset of the steady state data further corresponds to the particular network devices located in a same geographical location as the network device and of the same type of device;
• and
• perform one or more actions based on the one or more anomalies.

A non-transitory computer-readable medium storing a set of instructions, the set of instructions comprising:

• one or more instructions that, when executed by one or more processors of a device, cause the device to:
• receive key performance indicators (KPIs) corresponding to a plurality of network devices of a network;
• generate features based on the KPIs by generating the features corresponding to the KPIs related to components of the plurality of network devices, wherein the plurality of network devices include different types of network devices;
• select boundaries for a machine learning model based on the KPIs and the features;
• process the KPIs and the features, with the machine learning model and based on the boundaries, to generate steady state data corresponding to the plurality of network devices,
• wherein the steady state data is associated with a forecasted range for multiple percentiles in a future time related to the KPIs, that is determined based on using the machine learning model;
• receive new KPIs corresponding to a network device to be added to the network;
• generate new features based on the new KPIs;
• compare, during onboarding of the network device, the steady state data with the new KPIs and the new features to identify one or more anomalies corresponding to the network device being onboarded to be added to the network,
• wherein comparing the steady state data comprises:
• utilizing for the comparison, a subset of the steady state data corresponding to a subset of particular network devices of a same device type of the plurality of network devices as a type of device of the network device,
• wherein the subset of the steady state data further corresponds to the particular network devices located in a same geographical location as the network device and of the same type of device, and
• wherein the one or more anomalies correspond to one or more components of the network device;
• and
• perform one or more actions based on the one or more anomalies.