In code reviews and forum threads, one of the most common beginner questions is how to compare values correctly in Python. The confusion usually starts when people mix up equality vs identity, hit floating-point edge cases, or move from plain Python to libraries like NumPy or pandas where comparisons behave differently.<\/p>\n\n\n\n
I\u2019m new to Python and keep seeing two different ways to test inequality: Python gives you two main concepts here: value inequality and object identity inequality. Use them for different purposes.<\/p>\n\n\n\n If you manage images or videos programmatically, you can use these patterns to filter assets and enforce rules. For example, fetch assets and select those whose format is not equal to PNG, then transform or download them.<\/p>\n\n\n For quick one-off conversions or tests without writing code, explore Cloudinary Tools<\/a>.<\/p>\n\n\n\n!= and is not<\/code>. What is the correct way to check for not equal in Python, and what are the common pitfalls with numbers, strings, lists, None, and libraries like NumPy or pandas? Also, are there best practices for readability and performance?<\/em><\/p>\n\n\n\nAnswer:<\/h2>\n\n\n\n
1) The core operators<\/h3>\n\n\n\n
\n
a != b<\/code> checks if two values are not equal. This uses the same semantics as ==<\/code> and calls __eq__<\/code> under the hood.<\/li>\n\n\n\na is not b<\/code> checks if two references point to different objects in memory. This does not compare the value of the item, just whether or not they are the same<\/em> item.<\/li>\n<\/ul>\n\n\n# Value inequality<\/span>\n3<\/span> != 4<\/span>\u00a0 \u00a0 \u00a0 \u00a0 \u00a0 \u00a0 \u00a0 # True<\/span>\n[1<\/span>, 2<\/span>] != [1<\/span>, 2<\/span>]\u00a0 \u00a0 # False, lists have the same content<\/span>\n\n# Identity inequality<\/span>\na = [1<\/span>, 2<\/span>]\nb = [1<\/span>, 2<\/span>]\na is not b\u00a0 \u00a0 \u00a0 \u00a0 \u00a0 # True, different list objects even though values are equal<\/span><\/code><\/span>Code language:<\/span> PHP<\/span> (<\/span>php<\/span>)<\/span><\/small><\/pre>\n\n\n2) Best practices and common pitfalls<\/h3>\n\n\n\n
\n
if x is not None<\/code>. This is the canonical style and avoids surprises if __eq__<\/code> is overloaded.<\/li>\n\n\n\nmath.isclose<\/code> and invert the result:<\/li>\n<\/ul>\n\n\n# Value inequality<\/span>\n3<\/span> != 4<\/span>\u00a0 \u00a0 \u00a0 \u00a0 \u00a0 \u00a0 \u00a0 # True<\/span>\n[1<\/span>, 2<\/span>] != [1<\/span>, 2<\/span>]\u00a0 \u00a0 # False, lists have the same content<\/span>\n\n# Identity inequality<\/span>\na = [1<\/span>, 2<\/span>]\nb = [1<\/span>, 2<\/span>]\na is not b\u00a0 \u00a0 \u00a0 \u00a0 \u00a0 # True, different list objects even though values are equal<\/span><\/code><\/span>Code language:<\/span> PHP<\/span> (<\/span>php<\/span>)<\/span><\/small><\/pre>\n\n\n\n
float('nan') != float('nan')<\/code> is True by definition. Use math.isnan<\/code> to detect NaN.<\/li>\n\n\n\n\"Hello\" != \"hello<\/code>” is True. For case-insensitive checks, normalize:<\/li>\n<\/ul>\n\n\ns1.lower() != s2.lower()\n# or better for i18n<\/span>\ns1.casefold() != s2.casefold()<\/code><\/span>Code language:<\/span> PHP<\/span> (<\/span>php<\/span>)<\/span><\/small><\/pre>\n\n\n\n
__ne__<\/code> is automatically inferred as the inverse of __eq__<\/code> if you only define __eq__<\/code>. Dataclasses do this for you:<\/li>\n<\/ul>\n\n\nfrom<\/span> dataclasses<\/span> import<\/span> dataclass<\/span>\n\n@dataclass<\/span>(eq=True)\nclass User:<\/span>\n\u00a0 \u00a0 id:<\/span> int\n\u00a0 \u00a0 name:<\/span> str\n\nUser(1<\/span>, \"A\"<\/span>) != User(2<\/span>, \"A\"<\/span>)\u00a0 # True<\/code><\/span>Code language:<\/span> CSS<\/span> (<\/span>css<\/span>)<\/span><\/small><\/pre>\n\n\n\n
{'a'<\/span>: 1<\/span>} != {'a'<\/span>: 2<\/span>}\u00a0 # True<\/span>\n{1<\/span>, 2<\/span>} != {2<\/span>, 1<\/span>}\u00a0 \u00a0 # False, sets ignore order<\/span><\/code><\/span>Code language:<\/span> PHP<\/span> (<\/span>php<\/span>)<\/span><\/small><\/pre>\n\n\n\n
import numpy as<\/span> np\na = np.array<\/span>([1<\/span>, 2<\/span>, 3<\/span>])\nb = np.array<\/span>([1<\/span>, 9<\/span>, 3<\/span>])\n(a != b).any() \u00a0 # True if any element differs<\/span><\/code><\/span>Code language:<\/span> PHP<\/span> (<\/span>php<\/span>)<\/span><\/small><\/pre>\n\n\n\n
x != y != z<\/code> means (x != y)<\/code> and (y != z)<\/code>. It does not mean all three are distinct. For that, use len({x, y, z}) == 3<\/code>.<\/li>\n\n\n\nx != y<\/code> over not (x == y)<\/code>. The first is clearer and equivalent.<\/li>\n<\/ul>\n\n\n\n3) Quick reference examples<\/h3>\n\n\n
# Integers<\/span>\n5<\/span> != 5<\/span>\u00a0 \u00a0 \u00a0 \u00a0 \u00a0 # False<\/span>\n5<\/span> is not 5<\/span>\u00a0 \u00a0 \u00a0 # False for small ints due to interning, but don't rely on this<\/span>\n\n# None checks<\/span>\nvalue is not None\n\n# Strings<\/span>\n\"a\"<\/span> != \"A\"<\/span>\u00a0 \u00a0 \u00a0 # True<\/span>\n\"a\"<\/span>.casefold() != \"A\"<\/span>.casefold()\u00a0 # False<\/span>\n\n# Floats<\/span>\nfrom math import isclose, isnan\nnot isclose(0.30000000000000004<\/span>, 0.3<\/span>)\u00a0 # False<\/span>\nx = float('nan'<\/span>)\nx != x \u00a0 \u00a0 \u00a0 \u00a0 # True<\/span>\nisnan(x) \u00a0 \u00a0 \u00a0 # True<\/span>\n\n# Lists<\/span>\n[1<\/span>, 2<\/span>] != [2<\/span>, 1<\/span>]\u00a0 # True<\/span><\/code><\/span>Code language:<\/span> PHP<\/span> (<\/span>php<\/span>)<\/span><\/small><\/pre>\n\n\nApply this in a media pipeline with Cloudinary<\/h3>\n\n\n\n
# Example using Cloudinary's Python SDK concepts<\/span>\nimport cloudinary\nimport cloudinary.search\nimport cloudinary.uploader\n\ncloudinary.config(\n\u00a0 \u00a0 cloud_name=\"your_cloud\"<\/span>,\n\u00a0 \u00a0 api_key=\"your_key\"<\/span>,\n\u00a0 \u00a0 api_secret=\"your_secret\"<\/span>\n)\n\n# Fetch recent images<\/span>\nresults = cloudinary.search.Search() \\\n\u00a0 \u00a0 .expression(\"resource_type:image\"<\/span>) \\\n\u00a0 \u00a0 .max_results(100<\/span>) \\\n\u00a0 \u00a0 .execute()\n\n# Filter assets whose format is not equal to 'png'<\/span>\nnon_png_assets = [res for<\/span> res in results[\"resources\"<\/span>] if<\/span> res.get(\"format\"<\/span>) != \"png\"<\/span>]\n\n# Convert them on upload or via transformation if needed<\/span>\nfor<\/span> asset in non_png_assets:\n\u00a0 \u00a0 public_id = asset[\"public_id\"<\/span>]\n\u00a0 \u00a0 # Example: deliver as PNG by transformation, or re-upload as needed<\/span>\n\u00a0 \u00a0 # You can also automate conversions or use the Tools page for quick tests:<\/span>\n\u00a0 \u00a0 # https:\/\/cloudinary.com\/tools<\/span>\n\u00a0 \u00a0 # Delivered URL could be:<\/span>\n\u00a0 \u00a0 # https:\/\/res.cloudinary.com\/<cloud>\/image\/upload\/f_png\/<public_id>.jpg<\/span>\n\u00a0 \u00a0 pass<\/code><\/span>Code language:<\/span> PHP<\/span> (<\/span>php<\/span>)<\/span><\/small><\/pre>\n\n\nTL;DR<\/h3>\n\n\n\n
\n
!=<\/code> for value inequality, is not<\/code> for identity checks.<\/li>\n\n\n\nx is not None<\/code> when testing against None.<\/li>\n\n\n\nmath.isclose<\/code>; remember NaN is not equal to itself.<\/li>\n\n\n\nx != y != z<\/code> means all unique; use a set size check.<\/li>\n<\/ul>\n\n\n\nLearn More<\/h2>\n\n\n\n
\n